Skip to content

GitLab

"...gtest/git@developer.sourcefind.cn:yangql/googletest.git" did not exist on "5502540a5b5c5378824cd46591c2366bcf027555"
Commit 5b3e36dc authored by Sugon_ldc's avatar Sugon_ldc
Browse files

add model TSM

parents
Pipeline #315 failed with stages
in 0 seconds
Hide whitespace changes
Inline Side-by-side
Showing with 3196 additions and 0 deletions
configs/detection/acrn/slowfast_acrn_kinetics_pretrained_r50_8x8x1_cosine_10e_ava_rgb.py 0 → 100644
View file @ 5b3e36dc
model = dict(
type='FastRCNN',
backbone=dict(
type='ResNet3dSlowFast',
pretrained=None,
resample_rate=4,
speed_ratio=4,
channel_ratio=8,
slow_pathway=dict(
type='resnet3d',
depth=50,
pretrained=None,
lateral=True,
fusion_kernel=7,
conv1_kernel=(1, 7, 7),
dilations=(1, 1, 1, 1),
conv1_stride_t=1,
pool1_stride_t=1,
inflate=(0, 0, 1, 1),
spatial_strides=(1, 2, 2, 1)),
fast_pathway=dict(
type='resnet3d',
depth=50,
pretrained=None,
lateral=False,
base_channels=8,
conv1_kernel=(5, 7, 7),
conv1_stride_t=1,
pool1_stride_t=1,
spatial_strides=(1, 2, 2, 1))),
roi_head=dict(
type='AVARoIHead',
bbox_roi_extractor=dict(
type='SingleRoIExtractor3D',
roi_layer_type='RoIAlign',
output_size=8,
with_temporal_pool=True,
temporal_pool_mode='max'),
shared_head=dict(type='ACRNHead', in_channels=4608, out_channels=2304),
bbox_head=dict(
type='BBoxHeadAVA',
dropout_ratio=0.5,
in_channels=2304,
num_classes=81,
multilabel=True)),
train_cfg=dict(
rcnn=dict(
assigner=dict(
type='MaxIoUAssignerAVA',
pos_iou_thr=0.9,
neg_iou_thr=0.9,
min_pos_iou=0.9),
sampler=dict(
type='RandomSampler',
num=32,
pos_fraction=1,
neg_pos_ub=-1,
add_gt_as_proposals=True),
pos_weight=1.0,
debug=False)),
test_cfg=dict(rcnn=dict(action_thr=0.002)))
dataset_type = 'AVADataset'
data_root = 'data/ava/rawframes'
anno_root = 'data/ava/annotations'
ann_file_train = f'{anno_root}/ava_train_v2.1.csv'
ann_file_val = f'{anno_root}/ava_val_v2.1.csv'
exclude_file_train = f'{anno_root}/ava_train_excluded_timestamps_v2.1.csv'
exclude_file_val = f'{anno_root}/ava_val_excluded_timestamps_v2.1.csv'
label_file = f'{anno_root}/ava_action_list_v2.1_for_activitynet_2018.pbtxt'
proposal_file_train = (f'{anno_root}/ava_dense_proposals_train.FAIR.'
'recall_93.9.pkl')
proposal_file_val = f'{anno_root}/ava_dense_proposals_val.FAIR.recall_93.9.pkl'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
train_pipeline = [
dict(type='SampleAVAFrames', clip_len=32, frame_interval=2),
dict(type='RawFrameDecode'),
dict(type='RandomRescale', scale_range=(256, 320)),
dict(type='RandomCrop', size=256),
dict(type='Flip', flip_ratio=0.5),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals', 'gt_bboxes', 'gt_labels']),
dict(
type='ToDataContainer',
fields=[
dict(key=['proposals', 'gt_bboxes', 'gt_labels'], stack=False)
]),
dict(
type='Collect',
keys=['img', 'proposals', 'gt_bboxes', 'gt_labels'],
meta_keys=['scores', 'entity_ids'])
]
# The testing is w/o. any cropping / flipping
val_pipeline = [
dict(
type='SampleAVAFrames', clip_len=32, frame_interval=2, test_mode=True),
dict(type='RawFrameDecode'),
dict(type='Resize', scale=(-1, 256)),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals']),
dict(type='ToDataContainer', fields=[dict(key='proposals', stack=False)]),
dict(
type='Collect',
keys=['img', 'proposals'],
meta_keys=['scores', 'img_shape'],
nested=True)
]
data = dict(
videos_per_gpu=6,
workers_per_gpu=2,
val_dataloader=dict(videos_per_gpu=1),
test_dataloader=dict(videos_per_gpu=1),
train=dict(
type=dataset_type,
ann_file=ann_file_train,
exclude_file=exclude_file_train,
pipeline=train_pipeline,
label_file=label_file,
proposal_file=proposal_file_train,
person_det_score_thr=0.9,
data_prefix=data_root),
val=dict(
type=dataset_type,
ann_file=ann_file_val,
exclude_file=exclude_file_val,
pipeline=val_pipeline,
label_file=label_file,
proposal_file=proposal_file_val,
person_det_score_thr=0.9,
data_prefix=data_root))
data['test'] = data['val']
# optimizer
optimizer = dict(type='SGD', lr=0.075, momentum=0.9, weight_decay=0.00001)
# this lr is used for 8 gpus
optimizer_config = dict(grad_clip=dict(max_norm=40, norm_type=2))
# learning policy
lr_config = dict(
policy='CosineAnnealing',
by_epoch=False,
min_lr=0,
warmup='linear',
warmup_by_epoch=True,
warmup_iters=2,
warmup_ratio=0.1)
total_epochs = 10
checkpoint_config = dict(interval=1)
workflow = [('train', 1)]
evaluation = dict(interval=1)
log_config = dict(
interval=20, hooks=[
dict(type='TextLoggerHook'),
])
dist_params = dict(backend='nccl')
log_level = 'INFO'
work_dir = './work_dirs/slowfast_acrn_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb' # noqa: E501
load_from = 'https://download.openmmlab.com/mmaction/recognition/slowfast/slowfast_r50_8x8x1_256e_kinetics400_rgb/slowfast_r50_8x8x1_256e_kinetics400_rgb_20200716-73547d2b.pth' # noqa: E501
resume_from = None
find_unused_parameters = False
configs/detection/ava/README.md 0 → 100644
View file @ 5b3e36dc
# AVA
[Ava: A video dataset of spatio-temporally localized atomic visual actions](https://openaccess.thecvf.com/content_cvpr_2018/html/Gu_AVA_A_Video_CVPR_2018_paper.html)
<!-- [ALGORITHM] -->
<div align="center">
<img src="https://github.com/open-mmlab/mmaction2/raw/master/resources/spatio-temporal-det.gif" width="800px"/>
</div>
## Abstract
<!-- [ABSTRACT] -->
This paper introduces a video dataset of spatio-temporally localized Atomic Visual Actions (AVA). The AVA dataset densely annotates 80 atomic visual actions in 430 15-minute video clips, where actions are localized in space and time, resulting in 1.58M action labels with multiple labels per person occurring frequently. The key characteristics of our dataset are: (1) the definition of atomic visual actions, rather than composite actions; (2) precise spatio-temporal annotations with possibly multiple annotations for each person; (3) exhaustive annotation of these atomic actions over 15-minute video clips; (4) people temporally linked across consecutive segments; and (5) using movies to gather a varied set of action representations. This departs from existing datasets for spatio-temporal action recognition, which typically provide sparse annotations for composite actions in short video clips. We will release the dataset publicly.
AVA, with its realistic scene and action complexity, exposes the intrinsic difficulty of action recognition. To benchmark this, we present a novel approach for action localization that builds upon the current state-of-the-art methods, and demonstrates better performance on JHMDB and UCF101-24 categories. While setting a new state of the art on existing datasets, the overall results on AVA are low at 15.6% mAP, underscoring the need for developing new approaches for video understanding.
<!-- [IMAGE] -->
<div align=center>
<img src="https://user-images.githubusercontent.com/34324155/143015933-36eb7abd-d38f-4be6-a327-4d34c6f4edc1.png" width="800"/>
</div>
<!-- [ALGORITHM] -->
```BibTeX
@inproceedings{feichtenhofer2019slowfast,
title={Slowfast networks for video recognition},
author={Feichtenhofer, Christoph and Fan, Haoqi and Malik, Jitendra and He, Kaiming},
booktitle={Proceedings of the IEEE international conference on computer vision},
pages={6202--6211},
year={2019}
}
```
## Results and Models
### AVA2.1
| Model | Modality | Pretrained | Backbone | Input | gpus | Resolution | mAP | log | json | ckpt |
| :--------------------------------------------------------------------------------------------------------------------------------------------------: | :------: | :----------: | :-------: | :---: | :--: | :------------: | :---: | :-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | :-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | :---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
| [slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb](/configs/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py) | RGB | Kinetics-400 | ResNet50 | 4x16 | 8 | short-side 256 | 20.1 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201127.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201127.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217-40061d5f.pth) |
| [slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb](/configs/detection/ava/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb.py) | RGB | OmniSource | ResNet50 | 4x16 | 8 | short-side 256 | 21.8 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb_20201127.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb_20201127.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb_20201217-0c6d2e98.pth) |
| [slowonly_nl_kinetics_pretrained_r50_4x16x1_10e_ava_rgb](/configs/detection/ava/slowonly_nl_kinetics_pretrained_r50_4x16x1_10e_ava_rgb.py) | RGB | Kinetics-400 | ResNet50 | 4x16 | 8 | short-side 256 | 21.75 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowonly_nl_kinetics_pretrained_r50_4x16x1_10e_ava_rgb/20210316_122517.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowonly_nl_kinetics_pretrained_r50_4x16x1_10e_ava_rgb/20210316_122517.log.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowonly_nl_kinetics_pretrained_r50_4x16x1_10e_ava_rgb/slowonly_nl_kinetics_pretrained_r50_4x16x1_10e_ava_rgb_20210316-959829ec.pth) |
| [slowonly_nl_kinetics_pretrained_r50_8x8x1_10e_ava_rgb](/configs/detection/ava/slowonly_nl_kinetics_pretrained_r50_8x8x1_10e_ava_rgb.py) | RGB | Kinetics-400 | ResNet50 | 8x8 | 8x2 | short-side 256 | 23.79 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowonly_nl_kinetics_pretrained_r50_8x8x1_10e_ava_rgb/20210316_122517.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowonly_nl_kinetics_pretrained_r50_8x8x1_10e_ava_rgb/20210316_122517.log.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowonly_nl_kinetics_pretrained_r50_8x8x1_10e_ava_rgb/slowonly_nl_kinetics_pretrained_r50_8x8x1_10e_ava_rgb_20210316-5742e4dd.pth) |
| [slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb](/configs/detection/ava/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb.py) | RGB | Kinetics-400 | ResNet101 | 8x8 | 8x2 | short-side 256 | 24.6 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb_20201127.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb_20201127.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb_20201217-1c9b4117.pth) |
| [slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb](/configs/detection/ava/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb.py) | RGB | OmniSource | ResNet101 | 8x8 | 8x2 | short-side 256 | 25.9 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb_20201127.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb_20201127.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb_20201217-16378594.pth) |
| [slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb](/configs/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py) | RGB | Kinetics-400 | ResNet50 | 32x2 | 8x2 | short-side 256 | 24.4 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217-6e7c704d.pth) |
| [slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb](/configs/detection/ava/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py) | RGB | Kinetics-400 | ResNet50 | 32x2 | 8x2 | short-side 256 | 25.4 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201222.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201222.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201222-f4d209c9.pth) |
| [slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb](/configs/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb.py) | RGB | Kinetics-400 | ResNet50 | 32x2 | 8x2 | short-side 256 | 25.5 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb_20201217.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb_20201217.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb_20201217-ae225e97.pth) |
### AVA2.2
| Model | Modality | Pretrained | Backbone | Input | gpus | mAP | log | json | ckpt |
| :--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | :------: | :----------: | :------: | :---: | :--: | :--: | :--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | :----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
| [slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb](/configs/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.py) | RGB | Kinetics-400 | ResNet50 | 32x2 | 8 | 26.1 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb-b987b516.pth) |
| [slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb](/configs/detection/ava/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.py) | RGB | Kinetics-400 | ResNet50 | 32x2 | 8 | 26.4 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb-874e0845.pth) |
| [slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb](/configs/detection/ava/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.py) | RGB | Kinetics-400 | ResNet50 | 32x2 | 8 | 26.8 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb-345618cd.pth) |
:::{note}
1. The **gpus** indicates the number of gpu we used to get the checkpoint.
According to the [Linear Scaling Rule](https://arxiv.org/abs/1706.02677), you may set the learning rate proportional to the batch size if you use different GPUs or videos per GPU,
e.g., lr=0.01 for 4 GPUs x 2 video/gpu and lr=0.08 for 16 GPUs x 4 video/gpu.
2. **Context** indicates that using both RoI feature and global pooled feature for classification, which leads to around 1% mAP improvement in general.
:::
For more details on data preparation, you can refer to AVA in [Data Preparation](/docs/en/data_preparation.md).
## Train
You can use the following command to train a model.
```shell
python tools/train.py ${CONFIG_FILE} [optional arguments]
```
Example: train SlowOnly model on AVA with periodic validation.
```shell
python tools/train.py configs/detection/ava/slowonly_kinetics_pretrained_r50_8x8x1_20e_ava_rgb.py --validate
```
For more details and optional arguments infos, you can refer to **Training setting** part in [getting_started](/docs/en/getting_started.md#training-setting) .
### Train Custom Classes From Ava Dataset
You can train custom classes from ava. Ava suffers from class imbalance. There are more then 100,000 samples for classes like `stand`/`listen to (a person)`/`talk to (e.g., self, a person, a group)`/`watch (a person)`, whereas half of all classes has less than 500 samples. In most cases, training custom classes with fewer samples only will lead to better results.
Three steps to train custom classes:
- Step 1: Select custom classes from original classes, named `custom_classes`. Class `0` should not be selected since it is reserved for further usage (to identify whether a proposal is positive or negative, not implemented yet) and will be added automatically.
- Step 2: Set `num_classes`. In order to be compatible with current codes, Please make sure `num_classes == len(custom_classes) + 1`.
- The new class `0` corresponds to original class `0`. The new class `i`(i > 0) corresponds to original class `custom_classes[i-1]`.
- There are three `num_classes` in ava config, `model -> roi_head -> bbox_head -> num_classes`, `data -> train -> num_classes` and `data -> val -> num_classes`.
- If `num_classes <= 5`, input arg `topk` of `BBoxHeadAVA` should be modified. The default value of `topk` is `(3, 5)`, and all elements of `topk` must be smaller than `num_classes`.
- Step 3: Make sure all custom classes are in `label_file`. It is worth mentioning that there are two label files, `ava_action_list_v2.1_for_activitynet_2018.pbtxt`(contains 60 classes, 20 classes are missing) and `ava_action_list_v2.1.pbtxt`(contains all 80 classes).
Take `slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb` as an example, training custom classes with AP in range `(0.1, 0.3)`, aka `[3, 6, 10, 27, 29, 38, 41, 48, 51, 53, 54, 59, 61, 64, 70, 72]`. Please note that, the previously mentioned AP is calculated by original ckpt, which is trained by all 80 classes. The results are listed as follows.
| training classes | mAP(custom classes) | config | log | json | ckpt |
| :--------------: | :-----------------: | :----------------------------------------------------------------------------------------------------------------------------------------------------------------: | :-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | :---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | :-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
| All 80 classes | 0.1948 | [slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb](/configs/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py) | [log](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201127.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201127.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217-40061d5f.pth) |
| custom classes | 0.3311 | [slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes](/configs/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes.py) | [log](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes-4ab80419.pth) |
| All 80 classes | 0.1864 | [slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py](/configs/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py) | [log](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217-6e7c704d.pth) |
| custom classes | 0.3785 | [slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes](/configs/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes.py) | [log](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes_20210305.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes_20210305.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes_20210305-c6225546.pth) |
## Test
You can use the following command to test a model.
```shell
python tools/test.py ${CONFIG_FILE} ${CHECKPOINT_FILE} [optional arguments]
```
Example: test SlowOnly model on AVA and dump the result to a csv file.
```shell
python tools/test.py configs/detection/ava/slowonly_kinetics_pretrained_r50_8x8x1_20e_ava_rgb.py checkpoints/SOME_CHECKPOINT.pth --eval mAP --out results.csv
```
For more details and optional arguments infos, you can refer to **Test a dataset** part in [getting_started](/docs/en/getting_started.md#test-a-dataset) .
## Citation
<!-- [DATASET] -->
```BibTeX
@inproceedings{gu2018ava,
title={Ava: A video dataset of spatio-temporally localized atomic visual actions},
author={Gu, Chunhui and Sun, Chen and Ross, David A and Vondrick, Carl and Pantofaru, Caroline and Li, Yeqing and Vijayanarasimhan, Sudheendra and Toderici, George and Ricco, Susanna and Sukthankar, Rahul and others},
booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
pages={6047--6056},
year={2018}
}
```
```BibTeX
@article{duan2020omni,
title={Omni-sourced Webly-supervised Learning for Video Recognition},
author={Duan, Haodong and Zhao, Yue and Xiong, Yuanjun and Liu, Wentao and Lin, Dahua},
journal={arXiv preprint arXiv:2003.13042},
year={2020}
}
```
configs/detection/ava/README_zh-CN.md 0 → 100644
View file @ 5b3e36dc
# AVA
<div align="center">
<img src="https://github.com/open-mmlab/mmaction2/raw/master/resources/spatio-temporal-det.gif" width="800px"/>
</div>
## 简介
<!-- [DATASET] -->
```BibTeX
@inproceedings{gu2018ava,
title={Ava: A video dataset of spatio-temporally localized atomic visual actions},
author={Gu, Chunhui and Sun, Chen and Ross, David A and Vondrick, Carl and Pantofaru, Caroline and Li, Yeqing and Vijayanarasimhan, Sudheendra and Toderici, George and Ricco, Susanna and Sukthankar, Rahul and others},
booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
pages={6047--6056},
year={2018}
}
```
<!-- [ALGORITHM] -->
```BibTeX
@article{duan2020omni,
title={Omni-sourced Webly-supervised Learning for Video Recognition},
author={Duan, Haodong and Zhao, Yue and Xiong, Yuanjun and Liu, Wentao and Lin, Dahua},
journal={arXiv preprint arXiv:2003.13042},
year={2020}
}
```
<!-- [ALGORITHM] -->
```BibTeX
@inproceedings{feichtenhofer2019slowfast,
title={Slowfast networks for video recognition},
author={Feichtenhofer, Christoph and Fan, Haoqi and Malik, Jitendra and He, Kaiming},
booktitle={Proceedings of the IEEE international conference on computer vision},
pages={6202--6211},
year={2019}
}
```
## 模型库
### AVA2.1
| 配置文件 | 模态 | 预训练 | 主干网络 | 输入 | GPU 数量 | 分辨率 | mAP | log | json | ckpt |
| :--------------------------------------------------------------------------------------------------------------------------------------------------: | :--: | :----------: | :-------: | :--: | :------: | :------: | :---: | :-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | :-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | :---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
| [slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb](/configs/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py) | RGB | Kinetics-400 | ResNet50 | 4x16 | 8 | 短边 256 | 20.1 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201127.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201127.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217-40061d5f.pth) |
| [slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb](/configs/detection/ava/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb.py) | RGB | OmniSource | ResNet50 | 4x16 | 8 | 短边 256 | 21.8 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb_20201127.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb_20201127.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb_20201217-0c6d2e98.pth) |
| [slowonly_nl_kinetics_pretrained_r50_4x16x1_10e_ava_rgb](/configs/detection/ava/slowonly_nl_kinetics_pretrained_r50_4x16x1_10e_ava_rgb.py) | RGB | Kinetics-400 | ResNet50 | 4x16 | 8 | 短边 256 | 21.75 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowonly_nl_kinetics_pretrained_r50_4x16x1_10e_ava_rgb/20210316_122517.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowonly_nl_kinetics_pretrained_r50_4x16x1_10e_ava_rgb/20210316_122517.log.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowonly_nl_kinetics_pretrained_r50_4x16x1_10e_ava_rgb/slowonly_nl_kinetics_pretrained_r50_4x16x1_10e_ava_rgb_20210316-959829ec.pth) |
| [slowonly_nl_kinetics_pretrained_r50_8x8x1_10e_ava_rgb](/configs/detection/ava/slowonly_nl_kinetics_pretrained_r50_8x8x1_10e_ava_rgb.py) | RGB | Kinetics-400 | ResNet50 | 8x8 | 8x2 | 短边 256 | 23.79 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowonly_nl_kinetics_pretrained_r50_8x8x1_10e_ava_rgb/20210316_122517.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowonly_nl_kinetics_pretrained_r50_8x8x1_10e_ava_rgb/20210316_122517.log.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowonly_nl_kinetics_pretrained_r50_8x8x1_10e_ava_rgb/slowonly_nl_kinetics_pretrained_r50_8x8x1_10e_ava_rgb_20210316-5742e4dd.pth) |
| [slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb](/configs/detection/ava/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb.py) | RGB | Kinetics-400 | ResNet101 | 8x8 | 8x2 | 短边 256 | 24.6 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb_20201127.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb_20201127.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb_20201217-1c9b4117.pth) |
| [slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb](/configs/detection/ava/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb.py) | RGB | OmniSource | ResNet101 | 8x8 | 8x2 | 短边 256 | 25.9 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb_20201127.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb_20201127.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb_20201217-16378594.pth) |
| [slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb](/configs/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py) | RGB | Kinetics-400 | ResNet50 | 32x2 | 8x2 | 短边 256 | 24.4 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217-6e7c704d.pth) |
| [slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb](/configs/detection/ava/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py) | RGB | Kinetics-400 | ResNet50 | 32x2 | 8x2 | 短边 256 | 25.4 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201222.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201222.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201222-f4d209c9.pth) |
| [slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb](/configs/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb.py) | RGB | Kinetics-400 | ResNet50 | 32x2 | 8x2 | 短边 256 | 25.5 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb_20201217.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb_20201217.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb_20201217-ae225e97.pth) |
### AVA2.2
| 配置文件 | 模态 | 预训练 | 主干网络 | 输入 | GPU 数量 | mAP | log | json | ckpt |
| :--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | :--: | :----------: | :------: | :--: | :------: | :--: | :--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | :----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
| [slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb](/configs/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.py) | RGB | Kinetics-400 | ResNet50 | 32x2 | 8 | 26.1 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb-b987b516.pth) |
| [slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb](/configs/detection/ava/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.py) | RGB | Kinetics-400 | ResNet50 | 32x2 | 8 | 26.4 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb-874e0845.pth) |
| [slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb](/configs/detection/ava/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.py) | RGB | Kinetics-400 | ResNet50 | 32x2 | 8 | 26.8 | [log](https://download.openmmlab.com/mmaction/detection/ava/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb-345618cd.pth) |
注：
1. 这里的 **GPU 数量** 指的是得到模型权重文件对应的 GPU 个数。默认地，MMAction2 所提供的配置文件对应使用 8 块 GPU 进行训练的情况。
依据 [线性缩放规则](https://arxiv.org/abs/1706.02677)，当用户使用不同数量的 GPU 或者每块 GPU 处理不同视频个数时，需要根据批大小等比例地调节学习率。
如，lr=0.01 对应 4 GPUs x 2 video/gpu，以及 lr=0.08 对应 16 GPUs x 4 video/gpu。
2. **Context** 表示同时使用 RoI 特征与全局特征进行分类，可带来约 1% mAP 的提升。
对于数据集准备的细节，用户可参考 [数据准备](/docs/zh_cn/data_preparation.md)。
## 如何训练
用户可以使用以下指令进行模型训练。
```shell
python tools/train.py ${CONFIG_FILE} [optional arguments]
```
例如：在 AVA 数据集上训练 SlowOnly，并定期验证。
```shell
python tools/train.py configs/detection/ava/slowonly_kinetics_pretrained_r50_8x8x1_20e_ava_rgb.py --validate
```
更多训练细节，可参考 [基础教程](/docs/zh_cn/getting_started.md#训练配置) 中的 **训练配置** 部分。
### 训练 AVA 数据集中的自定义类别
用户可以训练 AVA 数据集中的自定义类别。AVA 中不同类别的样本量很不平衡：其中有超过 100000 样本的类别： `stand`/`listen to (a person)`/`talk to (e.g., self, a person, a group)`/`watch (a person)`，也有样本较少的类别（半数类别不足 500 样本）。大多数情况下，仅使用样本较少的类别进行训练将在这些类别上得到更好精度。
训练 AVA 数据集中的自定义类别包含 3 个步骤：
1. 从原先的类别中选择希望训练的类别，将其填写至配置文件的 `custom_classes` 域中。其中 `0` 不表示具体的动作类别，不应被选择。
2. 将 `num_classes` 设置为 `num_classes = len(custom_classes) + 1`。
- 在新的类别到编号的对应中，编号 `0` 仍对应原类别 `0`，编号 `i` (i > 0) 对应原类别 `custom_classes[i-1]`。
- 配置文件中 3 处涉及 `num_classes` 需要修改：`model -> roi_head -> bbox_head -> num_classes`， `data -> train -> num_classes`， `data -> val -> num_classes`.
- 若 `num_classes <= 5`， 配置文件 `BBoxHeadAVA` 中的 `topk` 参数应被修改。`topk` 的默认值为 `(3, 5)`，`topk` 中的所有元素应小于 `num_classes`。
3. 确认所有自定义类别在 `label_file` 中。
以 `slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb` 为例，这一配置文件训练所有 AP 在 `(0.1, 0.3)` 间的类别（这里的 AP 为 AVA 80 类训出模型的表现），即 `[3, 6, 10, 27, 29, 38, 41, 48, 51, 53, 54, 59, 61, 64, 70, 72]`。下表列出了自定义类别训练的模型精度：
| 训练类别 | mAP （自定义类别） | 配置文件 | log | json | ckpt |
| :--------: | :----------------: | :----------------------------------------------------------------------------------------------------------------------------------------------------------------: | :-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | :---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | :-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
| 全部 80 类 | 0.1948 | [slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb](/configs/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py) | [log](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201127.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201127.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217-40061d5f.pth) |
| 自定义类别 | 0.3311 | [slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes](/configs/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes.py) | [log](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes-4ab80419.pth) |
| 全部 80 类 | 0.1864 | [slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb](/configs/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py) | [log](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217-6e7c704d.pth) |
| 自定义类别 | 0.3785 | [slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes](/configs/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes.py) | [log](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes_20210305.log) | [json](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes_20210305.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes_20210305-c6225546.pth) |
## 如何测试
用户可以使用以下指令进行模型测试。
```shell
python tools/test.py ${CONFIG_FILE} ${CHECKPOINT_FILE} [optional arguments]
```
例如：在 AVA 上测试 SlowOnly 模型，并将结果存为 csv 文件。
```shell
python tools/test.py configs/detection/ava/slowonly_kinetics_pretrained_r50_8x8x1_20e_ava_rgb.py checkpoints/SOME_CHECKPOINT.pth --eval mAP --out results.csv
```
更多测试细节，可参考 [基础教程](/docs/zh_cn/getting_started.md#测试某个数据集) 中的 **测试某个数据集** 部分。
configs/detection/ava/metafile.yml 0 → 100644
View file @ 5b3e36dc
Collections:
- Name: AVA
README: configs/detection/ava/README.md
Paper:
URL: https://arxiv.org/abs/1705.08421
Title: "AVA: A Video Dataset of Spatio-temporally Localized Atomic Visual Actions"
Models:
- Config: configs/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py
In Collection: AVA
Metadata:
Architecture: ResNet50
Batch Size: 16
Epochs: 20
Input: 4x16
Pretrained: Kinetics-400
Resolution: short-side 256
Training Data: AVA v2.1
Training Resources: 8 GPUs
Modality: RGB
Name: slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb
Results:
- Dataset: AVA v2.1
Metrics:
mAP: 20.1
Task: Spatial Temporal Action Detection
Training Json Log: https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201127.json
Training Log: https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201127.log
Weights: https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217-40061d5f.pth
- Config: configs/detection/ava/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb.py
In Collection: AVA
Metadata:
Architecture: ResNet50
Batch Size: 16
Epochs: 20
Input: 4x16
Pretrained: OmniSource
Resolution: short-side 256
Training Data: AVA v2.1
Training Resources: 8 GPUs
Modality: RGB
Name: slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb
Results:
- Dataset: AVA v2.1
Metrics:
mAP: 21.8
Task: Spatial Temporal Action Detection
Training Json Log: https://download.openmmlab.com/mmaction/detection/ava/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb_20201127.json
Training Log: https://download.openmmlab.com/mmaction/detection/ava/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb_20201127.log
Weights: https://download.openmmlab.com/mmaction/detection/ava/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb_20201217-0c6d2e98.pth
- Config: configs/detection/ava/slowonly_nl_kinetics_pretrained_r50_4x16x1_10e_ava_rgb.py
In Collection: AVA
Metadata:
Architecture: ResNet50
Batch Size: 12
Epochs: 10
Input: 4x16
Pretrained: Kinetics-400
Resolution: short-side 256
Training Data: AVA v2.1
Training Resources: 8 GPUs
Modality: RGB
Name: slowonly_nl_kinetics_pretrained_r50_4x16x1_10e_ava_rgb
Results:
- Dataset: AVA v2.1
Metrics:
mAP: 21.75
Task: Spatial Temporal Action Detection
Training Json Log: https://download.openmmlab.com/mmaction/detection/ava/slowonly_nl_kinetics_pretrained_r50_4x16x1_10e_ava_rgb/20210316_122517.log.json
Training Log: https://download.openmmlab.com/mmaction/detection/ava/slowonly_nl_kinetics_pretrained_r50_4x16x1_10e_ava_rgb/20210316_122517.log
Weights: https://download.openmmlab.com/mmaction/detection/ava/slowonly_nl_kinetics_pretrained_r50_4x16x1_10e_ava_rgb/slowonly_nl_kinetics_pretrained_r50_4x16x1_10e_ava_rgb_20210316-959829ec.pth
- Config: configs/detection/ava/slowonly_nl_kinetics_pretrained_r50_8x8x1_10e_ava_rgb.py
In Collection: AVA
Metadata:
Architecture: ResNet50
Batch Size: 6
Epochs: 10
Input: 8x8
Pretrained: Kinetics-400
Resolution: short-side 256
Training Data: AVA v2.1
Training Resources: 16 GPUs
Modality: RGB
Name: slowonly_nl_kinetics_pretrained_r50_8x8x1_10e_ava_rgb
Results:
- Dataset: AVA v2.1
Metrics:
mAP: 23.79
Task: Spatial Temporal Action Detection
Training Json Log: https://download.openmmlab.com/mmaction/detection/ava/slowonly_nl_kinetics_pretrained_r50_8x8x1_10e_ava_rgb/20210316_122517.log.json
Training Log: https://download.openmmlab.com/mmaction/detection/ava/slowonly_nl_kinetics_pretrained_r50_8x8x1_10e_ava_rgb/20210316_122517.log
Weights: https://download.openmmlab.com/mmaction/detection/ava/slowonly_nl_kinetics_pretrained_r50_8x8x1_10e_ava_rgb/slowonly_nl_kinetics_pretrained_r50_8x8x1_10e_ava_rgb_20210316-5742e4dd.pth
- Config: configs/detection/ava/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb.py
In Collection: AVA
Metadata:
Architecture: ResNet101
Batch Size: 6
Epochs: 20
Input: 8x8
Pretrained: Kinetics-400
Resolution: short-side 256
Training Data: AVA v2.1
Training Resources: 16 GPUs
Modality: RGB
Name: slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb
Results:
- Dataset: AVA v2.1
Metrics:
mAP: 24.6
Task: Spatial Temporal Action Detection
Training Json Log: https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb_20201127.json
Training Log: https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb_20201127.log
Weights: https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb_20201217-1c9b4117.pth
- Config: configs/detection/ava/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb.py
In Collection: AVA
Metadata:
Architecture: ResNet101
Batch Size: 6
Epochs: 20
Input: 8x8
Pretrained: OmniSource
Resolution: short-side 256
Training Data: AVA v2.1
Training Resources: 16 GPUs
Modality: RGB
Name: slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb
Results:
- Dataset: AVA v2.1
Metrics:
mAP: 25.9
Task: Spatial Temporal Action Detection
Training Json Log: https://download.openmmlab.com/mmaction/detection/ava/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb_20201127.json
Training Log: https://download.openmmlab.com/mmaction/detection/ava/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb_20201127.log
Weights: https://download.openmmlab.com/mmaction/detection/ava/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb_20201217-16378594.pth
- Config: configs/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py
In Collection: AVA
Metadata:
Architecture: ResNet50
Batch Size: 9
Epochs: 20
Input: 32x2
Pretrained: Kinetics-400
Resolution: short-side 256
Training Data: AVA v2.1
Training Resources: 16 GPUs
Modality: RGB
Name: slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb
Results:
- Dataset: AVA v2.1
Metrics:
mAP: 24.4
Task: Spatial Temporal Action Detection
Training Json Log: https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217.json
Training Log: https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217.log
Weights: https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217-6e7c704d.pth
- Config: configs/detection/ava/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py
In Collection: AVA
Metadata:
Architecture: ResNet50
Batch Size: 9
Epochs: 20
Input: 32x2
Pretrained: Kinetics-400
Resolution: short-side 256
Training Data: AVA v2.1
Training Resources: 16 GPUs
Modality: RGB
Name: slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb
Results:
- Dataset: AVA v2.1
Metrics:
mAP: 25.4
Task: Spatial Temporal Action Detection
Training Json Log: https://download.openmmlab.com/mmaction/detection/ava/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201222.json
Training Log: https://download.openmmlab.com/mmaction/detection/ava/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201222.log
Weights: https://download.openmmlab.com/mmaction/detection/ava/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201222-f4d209c9.pth
- Config: configs/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb.py
In Collection: AVA
Metadata:
Architecture: ResNet50
Batch Size: 5
Epochs: 20
Input: 32x2
Pretrained: Kinetics-400
Resolution: short-side 256
Training Data: AVA v2.1
Training Resources: 16 GPUs
Modality: RGB
Name: slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb
Results:
- Dataset: AVA v2.1
Metrics:
mAP: 25.5
Task: Spatial Temporal Action Detection
Training Json Log: https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb_20201217.json
Training Log: https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb_20201217.log
Weights: https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb_20201217-ae225e97.pth
- Config: configs/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.py
In Collection: AVA
Metadata:
Architecture: ResNet50
Batch Size: 6
Epochs: 10
Input: 32x2
Pretrained: Kinetics-400
Resolution: short-side 256
Training Data: AVA v2.2
Training Resources: 8 GPUs
Modality: RGB
Name: slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb
Results:
- Dataset: AVA v2.2
Metrics:
mAP: 26.1
Task: Spatial Temporal Action Detection
Training Json Log: https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.json
Training Log: https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.log
Weights: https://download.openmmlab.com/mmaction/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb-b987b516.pth
- Config: configs/detection/ava/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.py
In Collection: AVA
Metadata:
Architecture: ResNet50
Batch Size: 6
Epochs: 10
Input: 32x2
Pretrained: Kinetics-400
Resolution: short-side 256
Training Data: AVA v2.2
Training Resources: 8 GPUs
Modality: RGB
Name: slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb
Results:
- Dataset: AVA v2.2
Metrics:
mAP: 26.8
Task: Spatial Temporal Action Detection
Training Json Log: https://download.openmmlab.com/mmaction/detection/ava/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.json
Training Log: https://download.openmmlab.com/mmaction/detection/ava/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.log
Weights: https://download.openmmlab.com/mmaction/detection/ava/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb-345618cd.pth
- Config: configs/detection/ava/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.py
In Collection: AVA
Metadata:
Architecture: ResNet50
Batch Size: 6
Epochs: 10
Input: 32x2
Pretrained: Kinetics-400
Resolution: short-side 256
Training Data: AVA v2.2
Training Resources: 8 GPUs
Modality: RGB
Name: slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb
Results:
- Dataset: AVA v2.2
Metrics:
mAP: 26.4
Task: Spatial Temporal Action Detection
Training Json Log: https://download.openmmlab.com/mmaction/detection/ava/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.json
Training Log: https://download.openmmlab.com/mmaction/detection/ava/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.log
Weights: https://download.openmmlab.com/mmaction/detection/ava/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb-874e0845.pth
configs/detection/ava/slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py 0 → 100644
View file @ 5b3e36dc
# model setting
model = dict(
type='FastRCNN',
backbone=dict(
type='ResNet3dSlowFast',
pretrained=None,
resample_rate=8,
speed_ratio=8,
channel_ratio=8,
slow_pathway=dict(
type='resnet3d',
depth=50,
pretrained=None,
lateral=True,
conv1_kernel=(1, 7, 7),
dilations=(1, 1, 1, 1),
conv1_stride_t=1,
pool1_stride_t=1,
inflate=(0, 0, 1, 1),
spatial_strides=(1, 2, 2, 1)),
fast_pathway=dict(
type='resnet3d',
depth=50,
pretrained=None,
lateral=False,
base_channels=8,
conv1_kernel=(5, 7, 7),
conv1_stride_t=1,
pool1_stride_t=1,
spatial_strides=(1, 2, 2, 1))),
roi_head=dict(
type='AVARoIHead',
bbox_roi_extractor=dict(
type='SingleRoIExtractor3D',
roi_layer_type='RoIAlign',
output_size=8,
with_temporal_pool=True,
with_global=True),
bbox_head=dict(
type='BBoxHeadAVA',
in_channels=4608,
num_classes=81,
multilabel=True,
dropout_ratio=0.5)),
train_cfg=dict(
rcnn=dict(
assigner=dict(
type='MaxIoUAssignerAVA',
pos_iou_thr=0.9,
neg_iou_thr=0.9,
min_pos_iou=0.9),
sampler=dict(
type='RandomSampler',
num=32,
pos_fraction=1,
neg_pos_ub=-1,
add_gt_as_proposals=True),
pos_weight=1.0,
debug=False)),
test_cfg=dict(rcnn=dict(action_thr=0.002)))
dataset_type = 'AVADataset'
data_root = 'data/ava/rawframes'
anno_root = 'data/ava/annotations'
ann_file_train = f'{anno_root}/ava_train_v2.1.csv'
ann_file_val = f'{anno_root}/ava_val_v2.1.csv'
exclude_file_train = f'{anno_root}/ava_train_excluded_timestamps_v2.1.csv'
exclude_file_val = f'{anno_root}/ava_val_excluded_timestamps_v2.1.csv'
label_file = f'{anno_root}/ava_action_list_v2.1_for_activitynet_2018.pbtxt'
proposal_file_train = (f'{anno_root}/ava_dense_proposals_train.FAIR.'
'recall_93.9.pkl')
proposal_file_val = f'{anno_root}/ava_dense_proposals_val.FAIR.recall_93.9.pkl'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
train_pipeline = [
dict(type='SampleAVAFrames', clip_len=32, frame_interval=2),
dict(type='RawFrameDecode'),
dict(type='RandomRescale', scale_range=(256, 320)),
dict(type='RandomCrop', size=256),
dict(type='Flip', flip_ratio=0.5),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals', 'gt_bboxes', 'gt_labels']),
dict(
type='ToDataContainer',
fields=[
dict(key=['proposals', 'gt_bboxes', 'gt_labels'], stack=False)
]),
dict(
type='Collect',
keys=['img', 'proposals', 'gt_bboxes', 'gt_labels'],
meta_keys=['scores', 'entity_ids'])
]
# The testing is w/o. any cropping / flipping
val_pipeline = [
dict(
type='SampleAVAFrames', clip_len=32, frame_interval=2, test_mode=True),
dict(type='RawFrameDecode'),
dict(type='Resize', scale=(-1, 256)),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals']),
dict(type='ToDataContainer', fields=[dict(key='proposals', stack=False)]),
dict(
type='Collect',
keys=['img', 'proposals'],
meta_keys=['scores', 'img_shape'],
nested=True)
]
data = dict(
videos_per_gpu=9,
workers_per_gpu=2,
val_dataloader=dict(videos_per_gpu=1),
test_dataloader=dict(videos_per_gpu=1),
train=dict(
type=dataset_type,
ann_file=ann_file_train,
exclude_file=exclude_file_train,
pipeline=train_pipeline,
label_file=label_file,
proposal_file=proposal_file_train,
person_det_score_thr=0.9,
data_prefix=data_root),
val=dict(
type=dataset_type,
ann_file=ann_file_val,
exclude_file=exclude_file_val,
pipeline=val_pipeline,
label_file=label_file,
proposal_file=proposal_file_val,
person_det_score_thr=0.9,
data_prefix=data_root))
data['test'] = data['val']
optimizer = dict(type='SGD', lr=0.1125, momentum=0.9, weight_decay=0.00001)
# this lr is used for 8 gpus
optimizer_config = dict(grad_clip=dict(max_norm=40, norm_type=2))
# learning policy
lr_config = dict(
policy='step',
step=[10, 15],
warmup='linear',
warmup_by_epoch=True,
warmup_iters=5,
warmup_ratio=0.1)
total_epochs = 20
checkpoint_config = dict(interval=1)
workflow = [('train', 1)]
evaluation = dict(interval=1, save_best='mAP@0.5IOU')
log_config = dict(
interval=20, hooks=[
dict(type='TextLoggerHook'),
])
dist_params = dict(backend='nccl')
log_level = 'INFO'
work_dir = ('./work_dirs/ava/'
'slowfast_context_kinetics_pretrained_r50_4x16x1_20e_ava_rgb')
load_from = ('https://download.openmmlab.com/mmaction/recognition/slowfast/'
'slowfast_r50_4x16x1_256e_kinetics400_rgb/'
'slowfast_r50_4x16x1_256e_kinetics400_rgb_20200704-bcde7ed7.pth')
resume_from = None
find_unused_parameters = False
configs/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py 0 → 100644
View file @ 5b3e36dc
# model setting
model = dict(
type='FastRCNN',
backbone=dict(
type='ResNet3dSlowFast',
pretrained=None,
resample_rate=8,
speed_ratio=8,
channel_ratio=8,
slow_pathway=dict(
type='resnet3d',
depth=50,
pretrained=None,
lateral=True,
conv1_kernel=(1, 7, 7),
dilations=(1, 1, 1, 1),
conv1_stride_t=1,
pool1_stride_t=1,
inflate=(0, 0, 1, 1),
spatial_strides=(1, 2, 2, 1)),
fast_pathway=dict(
type='resnet3d',
depth=50,
pretrained=None,
lateral=False,
base_channels=8,
conv1_kernel=(5, 7, 7),
conv1_stride_t=1,
pool1_stride_t=1,
spatial_strides=(1, 2, 2, 1))),
roi_head=dict(
type='AVARoIHead',
bbox_roi_extractor=dict(
type='SingleRoIExtractor3D',
roi_layer_type='RoIAlign',
output_size=8,
with_temporal_pool=True),
bbox_head=dict(
type='BBoxHeadAVA',
in_channels=2304,
num_classes=81,
multilabel=True,
dropout_ratio=0.5)),
train_cfg=dict(
rcnn=dict(
assigner=dict(
type='MaxIoUAssignerAVA',
pos_iou_thr=0.9,
neg_iou_thr=0.9,
min_pos_iou=0.9),
sampler=dict(
type='RandomSampler',
num=32,
pos_fraction=1,
neg_pos_ub=-1,
add_gt_as_proposals=True),
pos_weight=1.0,
debug=False)),
test_cfg=dict(rcnn=dict(action_thr=0.002)))
dataset_type = 'AVADataset'
data_root = 'data/ava/rawframes'
anno_root = 'data/ava/annotations'
ann_file_train = f'{anno_root}/ava_train_v2.1.csv'
ann_file_val = f'{anno_root}/ava_val_v2.1.csv'
exclude_file_train = f'{anno_root}/ava_train_excluded_timestamps_v2.1.csv'
exclude_file_val = f'{anno_root}/ava_val_excluded_timestamps_v2.1.csv'
label_file = f'{anno_root}/ava_action_list_v2.1_for_activitynet_2018.pbtxt'
proposal_file_train = (f'{anno_root}/ava_dense_proposals_train.FAIR.'
'recall_93.9.pkl')
proposal_file_val = f'{anno_root}/ava_dense_proposals_val.FAIR.recall_93.9.pkl'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
train_pipeline = [
dict(type='SampleAVAFrames', clip_len=32, frame_interval=2),
dict(type='RawFrameDecode'),
dict(type='RandomRescale', scale_range=(256, 320)),
dict(type='RandomCrop', size=256),
dict(type='Flip', flip_ratio=0.5),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals', 'gt_bboxes', 'gt_labels']),
dict(
type='ToDataContainer',
fields=[
dict(key=['proposals', 'gt_bboxes', 'gt_labels'], stack=False)
]),
dict(
type='Collect',
keys=['img', 'proposals', 'gt_bboxes', 'gt_labels'],
meta_keys=['scores', 'entity_ids'])
]
# The testing is w/o. any cropping / flipping
val_pipeline = [
dict(
type='SampleAVAFrames', clip_len=32, frame_interval=2, test_mode=True),
dict(type='RawFrameDecode'),
dict(type='Resize', scale=(-1, 256)),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals']),
dict(type='ToDataContainer', fields=[dict(key='proposals', stack=False)]),
dict(
type='Collect',
keys=['img', 'proposals'],
meta_keys=['scores', 'img_shape'],
nested=True)
]
data = dict(
videos_per_gpu=9,
workers_per_gpu=2,
val_dataloader=dict(videos_per_gpu=1),
test_dataloader=dict(videos_per_gpu=1),
train=dict(
type=dataset_type,
ann_file=ann_file_train,
exclude_file=exclude_file_train,
pipeline=train_pipeline,
label_file=label_file,
proposal_file=proposal_file_train,
person_det_score_thr=0.9,
data_prefix=data_root),
val=dict(
type=dataset_type,
ann_file=ann_file_val,
exclude_file=exclude_file_val,
pipeline=val_pipeline,
label_file=label_file,
proposal_file=proposal_file_val,
person_det_score_thr=0.9,
data_prefix=data_root))
data['test'] = data['val']
optimizer = dict(type='SGD', lr=0.1125, momentum=0.9, weight_decay=0.00001)
# this lr is used for 8 gpus
optimizer_config = dict(grad_clip=dict(max_norm=40, norm_type=2))
# learning policy
lr_config = dict(
policy='step',
step=[10, 15],
warmup='linear',
warmup_by_epoch=True,
warmup_iters=5,
warmup_ratio=0.1)
total_epochs = 20
checkpoint_config = dict(interval=1)
workflow = [('train', 1)]
evaluation = dict(interval=1, save_best='mAP@0.5IOU')
log_config = dict(
interval=20, hooks=[
dict(type='TextLoggerHook'),
])
dist_params = dict(backend='nccl')
log_level = 'INFO'
work_dir = ('./work_dirs/ava/'
'slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb')
load_from = ('https://download.openmmlab.com/mmaction/recognition/slowfast/'
'slowfast_r50_4x16x1_256e_kinetics400_rgb/'
'slowfast_r50_4x16x1_256e_kinetics400_rgb_20200704-bcde7ed7.pth')
resume_from = None
find_unused_parameters = False
configs/detection/ava/slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes.py 0 → 100644
View file @ 5b3e36dc
# custom classes of ava dataset
# Here we choose classes with AP in range [0.1, 0.3)
# AP is calculated by **slowonly** ckpt, which is trained by all 80 classes
custom_classes = [3, 6, 10, 27, 29, 38, 41, 48, 51, 53, 54, 59, 61, 64, 70, 72]
num_classes = len(custom_classes) + 1
# model setting
model = dict(
type='FastRCNN',
backbone=dict(
type='ResNet3dSlowFast',
pretrained=None,
resample_rate=8,
speed_ratio=8,
channel_ratio=8,
slow_pathway=dict(
type='resnet3d',
depth=50,
pretrained=None,
lateral=True,
conv1_kernel=(1, 7, 7),
dilations=(1, 1, 1, 1),
conv1_stride_t=1,
pool1_stride_t=1,
inflate=(0, 0, 1, 1),
spatial_strides=(1, 2, 2, 1)),
fast_pathway=dict(
type='resnet3d',
depth=50,
pretrained=None,
lateral=False,
base_channels=8,
conv1_kernel=(5, 7, 7),
conv1_stride_t=1,
pool1_stride_t=1,
spatial_strides=(1, 2, 2, 1))),
roi_head=dict(
type='AVARoIHead',
bbox_roi_extractor=dict(
type='SingleRoIExtractor3D',
roi_layer_type='RoIAlign',
output_size=8,
with_temporal_pool=True),
bbox_head=dict(
type='BBoxHeadAVA',
in_channels=2304,
num_classes=num_classes,
multilabel=True,
dropout_ratio=0.5)),
train_cfg=dict(
rcnn=dict(
assigner=dict(
type='MaxIoUAssignerAVA',
pos_iou_thr=0.9,
neg_iou_thr=0.9,
min_pos_iou=0.9),
sampler=dict(
type='RandomSampler',
num=32,
pos_fraction=1,
neg_pos_ub=-1,
add_gt_as_proposals=True),
pos_weight=1.0,
debug=False)),
test_cfg=dict(rcnn=dict(action_thr=0.002)))
dataset_type = 'AVADataset'
data_root = 'data/ava/rawframes'
anno_root = 'data/ava/annotations'
ann_file_train = f'{anno_root}/ava_train_v2.1.csv'
ann_file_val = f'{anno_root}/ava_val_v2.1.csv'
exclude_file_train = f'{anno_root}/ava_train_excluded_timestamps_v2.1.csv'
exclude_file_val = f'{anno_root}/ava_val_excluded_timestamps_v2.1.csv'
label_file = f'{anno_root}/ava_action_list_v2.1.pbtxt'
proposal_file_train = (f'{anno_root}/ava_dense_proposals_train.FAIR.'
'recall_93.9.pkl')
proposal_file_val = f'{anno_root}/ava_dense_proposals_val.FAIR.recall_93.9.pkl'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
train_pipeline = [
dict(type='SampleAVAFrames', clip_len=32, frame_interval=2),
dict(type='RawFrameDecode'),
dict(type='RandomRescale', scale_range=(256, 320)),
dict(type='RandomCrop', size=256),
dict(type='Flip', flip_ratio=0.5),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals', 'gt_bboxes', 'gt_labels']),
dict(
type='ToDataContainer',
fields=[
dict(key=['proposals', 'gt_bboxes', 'gt_labels'], stack=False)
]),
dict(
type='Collect',
keys=['img', 'proposals', 'gt_bboxes', 'gt_labels'],
meta_keys=['scores', 'entity_ids'])
]
# The testing is w/o. any cropping / flipping
val_pipeline = [
dict(
type='SampleAVAFrames', clip_len=32, frame_interval=2, test_mode=True),
dict(type='RawFrameDecode'),
dict(type='Resize', scale=(-1, 256)),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals']),
dict(type='ToDataContainer', fields=[dict(key='proposals', stack=False)]),
dict(
type='Collect',
keys=['img', 'proposals'],
meta_keys=['scores', 'img_shape'],
nested=True)
]
data = dict(
videos_per_gpu=9,
workers_per_gpu=2,
val_dataloader=dict(videos_per_gpu=1),
test_dataloader=dict(videos_per_gpu=1),
train=dict(
type=dataset_type,
ann_file=ann_file_train,
exclude_file=exclude_file_train,
pipeline=train_pipeline,
label_file=label_file,
proposal_file=proposal_file_train,
person_det_score_thr=0.9,
num_classes=num_classes,
custom_classes=custom_classes,
data_prefix=data_root),
val=dict(
type=dataset_type,
ann_file=ann_file_val,
exclude_file=exclude_file_val,
pipeline=val_pipeline,
label_file=label_file,
proposal_file=proposal_file_val,
person_det_score_thr=0.9,
num_classes=num_classes,
custom_classes=custom_classes,
data_prefix=data_root))
data['test'] = data['val']
optimizer = dict(type='SGD', lr=0.1125, momentum=0.9, weight_decay=0.00001)
# this lr is used for 8 gpus
optimizer_config = dict(grad_clip=dict(max_norm=40, norm_type=2))
# learning policy
lr_config = dict(
policy='step',
step=[10, 15],
warmup='linear',
warmup_by_epoch=True,
warmup_iters=5,
warmup_ratio=0.05)
total_epochs = 20
checkpoint_config = dict(interval=1)
workflow = [('train', 1)]
evaluation = dict(interval=1, save_best='mAP@0.5IOU')
log_config = dict(
interval=20, hooks=[
dict(type='TextLoggerHook'),
])
dist_params = dict(backend='nccl')
log_level = 'INFO'
work_dir = ('./work_dirs/ava/'
'slowfast_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom')
load_from = ('https://download.openmmlab.com/mmaction/recognition/slowfast/'
'slowfast_r50_4x16x1_256e_kinetics400_rgb/'
'slowfast_r50_4x16x1_256e_kinetics400_rgb_20200704-bcde7ed7.pth')
resume_from = None
find_unused_parameters = False
configs/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb.py 0 → 100644
View file @ 5b3e36dc
# model setting
model = dict(
type='FastRCNN',
backbone=dict(
type='ResNet3dSlowFast',
pretrained=None,
resample_rate=4,
speed_ratio=4,
channel_ratio=8,
slow_pathway=dict(
type='resnet3d',
depth=50,
pretrained=None,
lateral=True,
fusion_kernel=7,
conv1_kernel=(1, 7, 7),
dilations=(1, 1, 1, 1),
conv1_stride_t=1,
pool1_stride_t=1,
inflate=(0, 0, 1, 1),
spatial_strides=(1, 2, 2, 1)),
fast_pathway=dict(
type='resnet3d',
depth=50,
pretrained=None,
lateral=False,
base_channels=8,
conv1_kernel=(5, 7, 7),
conv1_stride_t=1,
pool1_stride_t=1,
spatial_strides=(1, 2, 2, 1))),
roi_head=dict(
type='AVARoIHead',
bbox_roi_extractor=dict(
type='SingleRoIExtractor3D',
roi_layer_type='RoIAlign',
output_size=8,
with_temporal_pool=True),
bbox_head=dict(
type='BBoxHeadAVA',
in_channels=2304,
num_classes=81,
multilabel=True,
dropout_ratio=0.5)),
train_cfg=dict(
rcnn=dict(
assigner=dict(
type='MaxIoUAssignerAVA',
pos_iou_thr=0.9,
neg_iou_thr=0.9,
min_pos_iou=0.9),
sampler=dict(
type='RandomSampler',
num=32,
pos_fraction=1,
neg_pos_ub=-1,
add_gt_as_proposals=True),
pos_weight=1.0,
debug=False)),
test_cfg=dict(rcnn=dict(action_thr=0.002)))
dataset_type = 'AVADataset'
data_root = 'data/ava/rawframes'
anno_root = 'data/ava/annotations'
ann_file_train = f'{anno_root}/ava_train_v2.1.csv'
ann_file_val = f'{anno_root}/ava_val_v2.1.csv'
exclude_file_train = f'{anno_root}/ava_train_excluded_timestamps_v2.1.csv'
exclude_file_val = f'{anno_root}/ava_val_excluded_timestamps_v2.1.csv'
label_file = f'{anno_root}/ava_action_list_v2.1_for_activitynet_2018.pbtxt'
proposal_file_train = (f'{anno_root}/ava_dense_proposals_train.FAIR.'
'recall_93.9.pkl')
proposal_file_val = f'{anno_root}/ava_dense_proposals_val.FAIR.recall_93.9.pkl'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
train_pipeline = [
dict(type='SampleAVAFrames', clip_len=32, frame_interval=2),
dict(type='RawFrameDecode'),
dict(type='RandomRescale', scale_range=(256, 320)),
dict(type='RandomCrop', size=256),
dict(type='Flip', flip_ratio=0.5),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals', 'gt_bboxes', 'gt_labels']),
dict(
type='ToDataContainer',
fields=[
dict(key=['proposals', 'gt_bboxes', 'gt_labels'], stack=False)
]),
dict(
type='Collect',
keys=['img', 'proposals', 'gt_bboxes', 'gt_labels'],
meta_keys=['scores', 'entity_ids'])
]
# The testing is w/o. any cropping / flipping
val_pipeline = [
dict(
type='SampleAVAFrames', clip_len=32, frame_interval=2, test_mode=True),
dict(type='RawFrameDecode'),
dict(type='Resize', scale=(-1, 256)),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals']),
dict(type='ToDataContainer', fields=[dict(key='proposals', stack=False)]),
dict(
type='Collect',
keys=['img', 'proposals'],
meta_keys=['scores', 'img_shape'],
nested=True)
]
data = dict(
videos_per_gpu=5,
workers_per_gpu=2,
val_dataloader=dict(videos_per_gpu=1),
test_dataloader=dict(videos_per_gpu=1),
train=dict(
type=dataset_type,
ann_file=ann_file_train,
exclude_file=exclude_file_train,
pipeline=train_pipeline,
label_file=label_file,
proposal_file=proposal_file_train,
person_det_score_thr=0.9,
data_prefix=data_root),
val=dict(
type=dataset_type,
ann_file=ann_file_val,
exclude_file=exclude_file_val,
pipeline=val_pipeline,
label_file=label_file,
proposal_file=proposal_file_val,
person_det_score_thr=0.9,
data_prefix=data_root))
data['test'] = data['val']
optimizer = dict(type='SGD', lr=0.075, momentum=0.9, weight_decay=0.00001)
# this lr is used for 8 gpus
optimizer_config = dict(grad_clip=dict(max_norm=40, norm_type=2))
# learning policy
lr_config = dict(
policy='step',
step=[10, 15],
warmup='linear',
warmup_by_epoch=True,
warmup_iters=5,
warmup_ratio=0.1)
total_epochs = 20
checkpoint_config = dict(interval=1)
workflow = [('train', 1)]
evaluation = dict(interval=1, save_best='mAP@0.5IOU')
log_config = dict(
interval=20, hooks=[
dict(type='TextLoggerHook'),
])
dist_params = dict(backend='nccl')
log_level = 'INFO'
work_dir = ('./work_dirs/ava/'
'slowfast_kinetics_pretrained_r50_8x8x1_20e_ava_rgb')
load_from = ('https://download.openmmlab.com/mmaction/recognition/slowfast/'
'slowfast_r50_8x8x1_256e_kinetics400_rgb/'
'slowfast_r50_8x8x1_256e_kinetics400_rgb_20200716-73547d2b.pth')
resume_from = None
find_unused_parameters = False
configs/detection/ava/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.py 0 → 100644
View file @ 5b3e36dc
model = dict(
type='FastRCNN',
backbone=dict(
type='ResNet3dSlowFast',
pretrained=None,
resample_rate=4,
speed_ratio=4,
channel_ratio=8,
slow_pathway=dict(
type='resnet3d',
depth=50,
pretrained=None,
lateral=True,
fusion_kernel=7,
conv1_kernel=(1, 7, 7),
dilations=(1, 1, 1, 1),
conv1_stride_t=1,
pool1_stride_t=1,
inflate=(0, 0, 1, 1),
spatial_strides=(1, 2, 2, 1)),
fast_pathway=dict(
type='resnet3d',
depth=50,
pretrained=None,
lateral=False,
base_channels=8,
conv1_kernel=(5, 7, 7),
conv1_stride_t=1,
pool1_stride_t=1,
spatial_strides=(1, 2, 2, 1))),
roi_head=dict(
type='AVARoIHead',
bbox_roi_extractor=dict(
type='SingleRoIExtractor3D',
roi_layer_type='RoIAlign',
output_size=8,
with_temporal_pool=True),
bbox_head=dict(
type='BBoxHeadAVA',
dropout_ratio=0.5,
in_channels=2304,
num_classes=81,
multilabel=True)),
train_cfg=dict(
rcnn=dict(
assigner=dict(
type='MaxIoUAssignerAVA',
pos_iou_thr=0.9,
neg_iou_thr=0.9,
min_pos_iou=0.9),
sampler=dict(
type='RandomSampler',
num=32,
pos_fraction=1,
neg_pos_ub=-1,
add_gt_as_proposals=True),
pos_weight=1.0,
debug=False)),
test_cfg=dict(rcnn=dict(action_thr=0.002)))
dataset_type = 'AVADataset'
data_root = 'data/ava/rawframes'
anno_root = 'data/ava/annotations'
ann_file_train = f'{anno_root}/ava_train_v2.2.csv'
ann_file_val = f'{anno_root}/ava_val_v2.2.csv'
exclude_file_train = f'{anno_root}/ava_train_excluded_timestamps_v2.2.csv'
exclude_file_val = f'{anno_root}/ava_val_excluded_timestamps_v2.2.csv'
label_file = f'{anno_root}/ava_action_list_v2.2_for_activitynet_2019.pbtxt'
proposal_file_train = (f'{anno_root}/ava_dense_proposals_train.FAIR.'
'recall_93.9.pkl')
proposal_file_val = f'{anno_root}/ava_dense_proposals_val.FAIR.recall_93.9.pkl'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
train_pipeline = [
dict(type='SampleAVAFrames', clip_len=32, frame_interval=2),
dict(type='RawFrameDecode'),
dict(type='RandomRescale', scale_range=(256, 320)),
dict(type='RandomCrop', size=256),
dict(type='Flip', flip_ratio=0.5),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals', 'gt_bboxes', 'gt_labels']),
dict(
type='ToDataContainer',
fields=[
dict(key=['proposals', 'gt_bboxes', 'gt_labels'], stack=False)
]),
dict(
type='Collect',
keys=['img', 'proposals', 'gt_bboxes', 'gt_labels'],
meta_keys=['scores', 'entity_ids'])
]
# The testing is w/o. any cropping / flipping
val_pipeline = [
dict(
type='SampleAVAFrames', clip_len=32, frame_interval=2, test_mode=True),
dict(type='RawFrameDecode'),
dict(type='Resize', scale=(-1, 256)),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals']),
dict(type='ToDataContainer', fields=[dict(key='proposals', stack=False)]),
dict(
type='Collect',
keys=['img', 'proposals'],
meta_keys=['scores', 'img_shape'],
nested=True)
]
data = dict(
videos_per_gpu=6,
workers_per_gpu=2,
val_dataloader=dict(videos_per_gpu=1),
test_dataloader=dict(videos_per_gpu=1),
train=dict(
type=dataset_type,
ann_file=ann_file_train,
exclude_file=exclude_file_train,
pipeline=train_pipeline,
label_file=label_file,
proposal_file=proposal_file_train,
person_det_score_thr=0.9,
data_prefix=data_root),
val=dict(
type=dataset_type,
ann_file=ann_file_val,
exclude_file=exclude_file_val,
pipeline=val_pipeline,
label_file=label_file,
proposal_file=proposal_file_val,
person_det_score_thr=0.9,
data_prefix=data_root))
data['test'] = data['val']
# optimizer
optimizer = dict(type='SGD', lr=0.075, momentum=0.9, weight_decay=0.00001)
# this lr is used for 8 gpus
optimizer_config = dict(grad_clip=dict(max_norm=40, norm_type=2))
# learning policy
lr_config = dict(
policy='CosineAnnealing',
by_epoch=False,
min_lr=0,
warmup='linear',
warmup_by_epoch=True,
warmup_iters=2,
warmup_ratio=0.1)
total_epochs = 10
checkpoint_config = dict(interval=1)
workflow = [('train', 1)]
evaluation = dict(interval=1)
log_config = dict(
interval=20, hooks=[
dict(type='TextLoggerHook'),
])
dist_params = dict(backend='nccl')
log_level = 'INFO'
work_dir = './work_dirs/slowfast_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb' # noqa: E501
load_from = 'https://download.openmmlab.com/mmaction/recognition/slowfast/slowfast_r50_8x8x1_256e_kinetics400_rgb/slowfast_r50_8x8x1_256e_kinetics400_rgb_20200716-73547d2b.pth' # noqa: E501
resume_from = None
find_unused_parameters = False
configs/detection/ava/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.py 0 → 100644
View file @ 5b3e36dc
model = dict(
type='FastRCNN',
backbone=dict(
type='ResNet3dSlowFast',
pretrained=None,
resample_rate=4,
speed_ratio=4,
channel_ratio=8,
slow_pathway=dict(
type='resnet3d',
depth=50,
pretrained=None,
lateral=True,
fusion_kernel=7,
conv1_kernel=(1, 7, 7),
dilations=(1, 1, 1, 1),
conv1_stride_t=1,
pool1_stride_t=1,
inflate=(0, 0, 1, 1),
spatial_strides=(1, 2, 2, 1)),
fast_pathway=dict(
type='resnet3d',
depth=50,
pretrained=None,
lateral=False,
base_channels=8,
conv1_kernel=(5, 7, 7),
conv1_stride_t=1,
pool1_stride_t=1,
spatial_strides=(1, 2, 2, 1))),
roi_head=dict(
type='AVARoIHead',
bbox_roi_extractor=dict(
type='SingleRoIExtractor3D',
roi_layer_type='RoIAlign',
output_size=8,
with_temporal_pool=True,
temporal_pool_mode='max'),
bbox_head=dict(
type='BBoxHeadAVA',
dropout_ratio=0.5,
in_channels=2304,
focal_alpha=3.0,
focal_gamma=1.0,
num_classes=81,
multilabel=True)),
train_cfg=dict(
rcnn=dict(
assigner=dict(
type='MaxIoUAssignerAVA',
pos_iou_thr=0.9,
neg_iou_thr=0.9,
min_pos_iou=0.9),
sampler=dict(
type='RandomSampler',
num=32,
pos_fraction=1,
neg_pos_ub=-1,
add_gt_as_proposals=True),
pos_weight=1.0,
debug=False)),
test_cfg=dict(rcnn=dict(action_thr=0.002)))
dataset_type = 'AVADataset'
data_root = 'data/ava/rawframes'
anno_root = 'data/ava/annotations'
ann_file_train = f'{anno_root}/ava_train_v2.2.csv'
ann_file_val = f'{anno_root}/ava_val_v2.2.csv'
exclude_file_train = f'{anno_root}/ava_train_excluded_timestamps_v2.2.csv'
exclude_file_val = f'{anno_root}/ava_val_excluded_timestamps_v2.2.csv'
label_file = f'{anno_root}/ava_action_list_v2.2_for_activitynet_2019.pbtxt'
proposal_file_train = (f'{anno_root}/ava_dense_proposals_train.FAIR.'
'recall_93.9.pkl')
proposal_file_val = f'{anno_root}/ava_dense_proposals_val.FAIR.recall_93.9.pkl'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
train_pipeline = [
dict(type='SampleAVAFrames', clip_len=32, frame_interval=2),
dict(type='RawFrameDecode'),
dict(type='RandomRescale', scale_range=(256, 320)),
dict(type='RandomCrop', size=256),
dict(type='Flip', flip_ratio=0.5),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals', 'gt_bboxes', 'gt_labels']),
dict(
type='ToDataContainer',
fields=[
dict(key=['proposals', 'gt_bboxes', 'gt_labels'], stack=False)
]),
dict(
type='Collect',
keys=['img', 'proposals', 'gt_bboxes', 'gt_labels'],
meta_keys=['scores', 'entity_ids'])
]
# The testing is w/o. any cropping / flipping
val_pipeline = [
dict(
type='SampleAVAFrames', clip_len=32, frame_interval=2, test_mode=True),
dict(type='RawFrameDecode'),
dict(type='Resize', scale=(-1, 256)),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals']),
dict(type='ToDataContainer', fields=[dict(key='proposals', stack=False)]),
dict(
type='Collect',
keys=['img', 'proposals'],
meta_keys=['scores', 'img_shape'],
nested=True)
]
data = dict(
videos_per_gpu=6,
workers_per_gpu=2,
val_dataloader=dict(videos_per_gpu=1),
test_dataloader=dict(videos_per_gpu=1),
train=dict(
type=dataset_type,
ann_file=ann_file_train,
exclude_file=exclude_file_train,
pipeline=train_pipeline,
label_file=label_file,
proposal_file=proposal_file_train,
person_det_score_thr=0.9,
data_prefix=data_root),
val=dict(
type=dataset_type,
ann_file=ann_file_val,
exclude_file=exclude_file_val,
pipeline=val_pipeline,
label_file=label_file,
proposal_file=proposal_file_val,
person_det_score_thr=0.9,
data_prefix=data_root))
data['test'] = data['val']
# optimizer
optimizer = dict(type='SGD', lr=0.075, momentum=0.9, weight_decay=0.00001)
# this lr is used for 8 gpus
optimizer_config = dict(grad_clip=dict(max_norm=40, norm_type=2))
# learning policy
lr_config = dict(
policy='CosineAnnealing',
by_epoch=False,
min_lr=0,
warmup='linear',
warmup_by_epoch=True,
warmup_iters=2,
warmup_ratio=0.1)
total_epochs = 10
checkpoint_config = dict(interval=1)
workflow = [('train', 1)]
evaluation = dict(interval=1)
log_config = dict(
interval=20, hooks=[
dict(type='TextLoggerHook'),
])
dist_params = dict(backend='nccl')
log_level = 'INFO'
work_dir = './work_dirs/slowfast_temporal_max_focal_alpha3_gamma1_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb' # noqa: E501
load_from = 'https://download.openmmlab.com/mmaction/recognition/slowfast/slowfast_r50_8x8x1_256e_kinetics400_rgb/slowfast_r50_8x8x1_256e_kinetics400_rgb_20200716-73547d2b.pth' # noqa: E501
resume_from = None
find_unused_parameters = False
configs/detection/ava/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb.py 0 → 100644
View file @ 5b3e36dc
model = dict(
type='FastRCNN',
backbone=dict(
type='ResNet3dSlowFast',
pretrained=None,
resample_rate=4,
speed_ratio=4,
channel_ratio=8,
slow_pathway=dict(
type='resnet3d',
depth=50,
pretrained=None,
lateral=True,
fusion_kernel=7,
conv1_kernel=(1, 7, 7),
dilations=(1, 1, 1, 1),
conv1_stride_t=1,
pool1_stride_t=1,
inflate=(0, 0, 1, 1),
spatial_strides=(1, 2, 2, 1)),
fast_pathway=dict(
type='resnet3d',
depth=50,
pretrained=None,
lateral=False,
base_channels=8,
conv1_kernel=(5, 7, 7),
conv1_stride_t=1,
pool1_stride_t=1,
spatial_strides=(1, 2, 2, 1))),
roi_head=dict(
type='AVARoIHead',
bbox_roi_extractor=dict(
type='SingleRoIExtractor3D',
roi_layer_type='RoIAlign',
output_size=8,
with_temporal_pool=True,
temporal_pool_mode='max'),
bbox_head=dict(
type='BBoxHeadAVA',
dropout_ratio=0.5,
in_channels=2304,
num_classes=81,
multilabel=True)),
train_cfg=dict(
rcnn=dict(
assigner=dict(
type='MaxIoUAssignerAVA',
pos_iou_thr=0.9,
neg_iou_thr=0.9,
min_pos_iou=0.9),
sampler=dict(
type='RandomSampler',
num=32,
pos_fraction=1,
neg_pos_ub=-1,
add_gt_as_proposals=True),
pos_weight=1.0,
debug=False)),
test_cfg=dict(rcnn=dict(action_thr=0.002)))
dataset_type = 'AVADataset'
data_root = 'data/ava/rawframes'
anno_root = 'data/ava/annotations'
ann_file_train = f'{anno_root}/ava_train_v2.2.csv'
ann_file_val = f'{anno_root}/ava_val_v2.2.csv'
exclude_file_train = f'{anno_root}/ava_train_excluded_timestamps_v2.2.csv'
exclude_file_val = f'{anno_root}/ava_val_excluded_timestamps_v2.2.csv'
label_file = f'{anno_root}/ava_action_list_v2.2_for_activitynet_2019.pbtxt'
proposal_file_train = (f'{anno_root}/ava_dense_proposals_train.FAIR.'
'recall_93.9.pkl')
proposal_file_val = f'{anno_root}/ava_dense_proposals_val.FAIR.recall_93.9.pkl'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
train_pipeline = [
dict(type='SampleAVAFrames', clip_len=32, frame_interval=2),
dict(type='RawFrameDecode'),
dict(type='RandomRescale', scale_range=(256, 320)),
dict(type='RandomCrop', size=256),
dict(type='Flip', flip_ratio=0.5),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals', 'gt_bboxes', 'gt_labels']),
dict(
type='ToDataContainer',
fields=[
dict(key=['proposals', 'gt_bboxes', 'gt_labels'], stack=False)
]),
dict(
type='Collect',
keys=['img', 'proposals', 'gt_bboxes', 'gt_labels'],
meta_keys=['scores', 'entity_ids'])
]
# The testing is w/o. any cropping / flipping
val_pipeline = [
dict(
type='SampleAVAFrames', clip_len=32, frame_interval=2, test_mode=True),
dict(type='RawFrameDecode'),
dict(type='Resize', scale=(-1, 256)),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals']),
dict(type='ToDataContainer', fields=[dict(key='proposals', stack=False)]),
dict(
type='Collect',
keys=['img', 'proposals'],
meta_keys=['scores', 'img_shape'],
nested=True)
]
data = dict(
videos_per_gpu=6,
workers_per_gpu=2,
val_dataloader=dict(videos_per_gpu=1),
test_dataloader=dict(videos_per_gpu=1),
train=dict(
type=dataset_type,
ann_file=ann_file_train,
exclude_file=exclude_file_train,
pipeline=train_pipeline,
label_file=label_file,
proposal_file=proposal_file_train,
person_det_score_thr=0.9,
data_prefix=data_root),
val=dict(
type=dataset_type,
ann_file=ann_file_val,
exclude_file=exclude_file_val,
pipeline=val_pipeline,
label_file=label_file,
proposal_file=proposal_file_val,
person_det_score_thr=0.9,
data_prefix=data_root))
data['test'] = data['val']
# optimizer
optimizer = dict(type='SGD', lr=0.075, momentum=0.9, weight_decay=0.00001)
# this lr is used for 8 gpus
optimizer_config = dict(grad_clip=dict(max_norm=40, norm_type=2))
# learning policy
lr_config = dict(
policy='CosineAnnealing',
by_epoch=False,
min_lr=0,
warmup='linear',
warmup_by_epoch=True,
warmup_iters=2,
warmup_ratio=0.1)
total_epochs = 10
checkpoint_config = dict(interval=1)
workflow = [('train', 1)]
evaluation = dict(interval=1)
log_config = dict(
interval=20, hooks=[
dict(type='TextLoggerHook'),
])
dist_params = dict(backend='nccl')
log_level = 'INFO'
work_dir = './work_dirs/slowfast_temporal_max_kinetics_pretrained_r50_8x8x1_cosine_10e_ava22_rgb' # noqa: E501
load_from = 'https://download.openmmlab.com/mmaction/recognition/slowfast/slowfast_r50_8x8x1_256e_kinetics400_rgb/slowfast_r50_8x8x1_256e_kinetics400_rgb_20200716-73547d2b.pth' # noqa: E501
resume_from = None
find_unused_parameters = False
configs/detection/ava/slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb.py 0 → 100644
View file @ 5b3e36dc
# model setting
model = dict(
type='FastRCNN',
backbone=dict(
type='ResNet3dSlowOnly',
depth=101,
pretrained=None,
pretrained2d=False,
lateral=False,
num_stages=4,
conv1_kernel=(1, 7, 7),
conv1_stride_t=1,
pool1_stride_t=1,
spatial_strides=(1, 2, 2, 1)),
roi_head=dict(
type='AVARoIHead',
bbox_roi_extractor=dict(
type='SingleRoIExtractor3D',
roi_layer_type='RoIAlign',
output_size=8,
with_temporal_pool=True),
bbox_head=dict(
type='BBoxHeadAVA',
in_channels=2048,
num_classes=81,
multilabel=True,
dropout_ratio=0.5)),
train_cfg=dict(
rcnn=dict(
assigner=dict(
type='MaxIoUAssignerAVA',
pos_iou_thr=0.9,
neg_iou_thr=0.9,
min_pos_iou=0.9),
sampler=dict(
type='RandomSampler',
num=32,
pos_fraction=1,
neg_pos_ub=-1,
add_gt_as_proposals=True),
pos_weight=1.0,
debug=False)),
test_cfg=dict(rcnn=dict(action_thr=0.002)))
dataset_type = 'AVADataset'
data_root = 'data/ava/rawframes'
anno_root = 'data/ava/annotations'
ann_file_train = f'{anno_root}/ava_train_v2.1.csv'
ann_file_val = f'{anno_root}/ava_val_v2.1.csv'
exclude_file_train = f'{anno_root}/ava_train_excluded_timestamps_v2.1.csv'
exclude_file_val = f'{anno_root}/ava_val_excluded_timestamps_v2.1.csv'
label_file = f'{anno_root}/ava_action_list_v2.1_for_activitynet_2018.pbtxt'
proposal_file_train = (f'{anno_root}/ava_dense_proposals_train.FAIR.'
'recall_93.9.pkl')
proposal_file_val = f'{anno_root}/ava_dense_proposals_val.FAIR.recall_93.9.pkl'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
train_pipeline = [
dict(type='SampleAVAFrames', clip_len=8, frame_interval=8),
dict(type='RawFrameDecode'),
dict(type='RandomRescale', scale_range=(256, 320)),
dict(type='RandomCrop', size=256),
dict(type='Flip', flip_ratio=0.5),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals', 'gt_bboxes', 'gt_labels']),
dict(
type='ToDataContainer',
fields=[
dict(key=['proposals', 'gt_bboxes', 'gt_labels'], stack=False)
]),
dict(
type='Collect',
keys=['img', 'proposals', 'gt_bboxes', 'gt_labels'],
meta_keys=['scores', 'entity_ids'])
]
# The testing is w/o. any cropping / flipping
val_pipeline = [
dict(type='SampleAVAFrames', clip_len=8, frame_interval=8, test_mode=True),
dict(type='RawFrameDecode'),
dict(type='Resize', scale=(-1, 256)),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals']),
dict(type='ToDataContainer', fields=[dict(key='proposals', stack=False)]),
dict(
type='Collect',
keys=['img', 'proposals'],
meta_keys=['scores', 'img_shape'],
nested=True)
]
data = dict(
videos_per_gpu=6,
workers_per_gpu=2,
# During testing, each video may have different shape
val_dataloader=dict(videos_per_gpu=1),
test_dataloader=dict(videos_per_gpu=1),
train=dict(
type=dataset_type,
ann_file=ann_file_train,
exclude_file=exclude_file_train,
pipeline=train_pipeline,
label_file=label_file,
proposal_file=proposal_file_train,
person_det_score_thr=0.9,
data_prefix=data_root),
val=dict(
type=dataset_type,
ann_file=ann_file_val,
exclude_file=exclude_file_val,
pipeline=val_pipeline,
label_file=label_file,
proposal_file=proposal_file_val,
person_det_score_thr=0.9,
data_prefix=data_root))
data['test'] = data['val']
optimizer = dict(type='SGD', lr=0.075, momentum=0.9, weight_decay=0.00001)
# this lr is used for 8 gpus
optimizer_config = dict(grad_clip=dict(max_norm=40, norm_type=2))
# learning policy
lr_config = dict(
policy='step',
step=[10, 15],
warmup='linear',
warmup_by_epoch=True,
warmup_iters=5,
warmup_ratio=0.1)
total_epochs = 20
checkpoint_config = dict(interval=1)
workflow = [('train', 1)]
evaluation = dict(interval=1, save_best='mAP@0.5IOU')
log_config = dict(
interval=20, hooks=[
dict(type='TextLoggerHook'),
])
dist_params = dict(backend='nccl')
log_level = 'INFO'
work_dir = ('./work_dirs/ava/'
'slowonly_kinetics_pretrained_r101_8x8x1_20e_ava_rgb')
load_from = ('https://download.openmmlab.com/mmaction/recognition/slowonly/'
'omni/slowonly_r101_without_omni_8x8x1_'
'kinetics400_rgb_20200926-0c730aef.pth')
resume_from = None
find_unused_parameters = False
configs/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py 0 → 100644
View file @ 5b3e36dc
# model setting
model = dict(
type='FastRCNN',
backbone=dict(
type='ResNet3dSlowOnly',
depth=50,
pretrained=None,
pretrained2d=False,
lateral=False,
num_stages=4,
conv1_kernel=(1, 7, 7),
conv1_stride_t=1,
pool1_stride_t=1,
spatial_strides=(1, 2, 2, 1)),
roi_head=dict(
type='AVARoIHead',
bbox_roi_extractor=dict(
type='SingleRoIExtractor3D',
roi_layer_type='RoIAlign',
output_size=8,
with_temporal_pool=True),
bbox_head=dict(
type='BBoxHeadAVA',
in_channels=2048,
num_classes=81,
multilabel=True,
dropout_ratio=0.5)),
train_cfg=dict(
rcnn=dict(
assigner=dict(
type='MaxIoUAssignerAVA',
pos_iou_thr=0.9,
neg_iou_thr=0.9,
min_pos_iou=0.9),
sampler=dict(
type='RandomSampler',
num=32,
pos_fraction=1,
neg_pos_ub=-1,
add_gt_as_proposals=True),
pos_weight=1.0,
debug=False)),
test_cfg=dict(rcnn=dict(action_thr=0.002)))
dataset_type = 'AVADataset'
data_root = 'data/ava/rawframes'
anno_root = 'data/ava/annotations'
ann_file_train = f'{anno_root}/ava_train_v2.1.csv'
ann_file_val = f'{anno_root}/ava_val_v2.1.csv'
exclude_file_train = f'{anno_root}/ava_train_excluded_timestamps_v2.1.csv'
exclude_file_val = f'{anno_root}/ava_val_excluded_timestamps_v2.1.csv'
label_file = f'{anno_root}/ava_action_list_v2.1_for_activitynet_2018.pbtxt'
proposal_file_train = (f'{anno_root}/ava_dense_proposals_train.FAIR.'
'recall_93.9.pkl')
proposal_file_val = f'{anno_root}/ava_dense_proposals_val.FAIR.recall_93.9.pkl'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
train_pipeline = [
dict(type='SampleAVAFrames', clip_len=4, frame_interval=16),
dict(type='RawFrameDecode'),
dict(type='RandomRescale', scale_range=(256, 320)),
dict(type='RandomCrop', size=256),
dict(type='Flip', flip_ratio=0.5),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals', 'gt_bboxes', 'gt_labels']),
dict(
type='ToDataContainer',
fields=[
dict(key=['proposals', 'gt_bboxes', 'gt_labels'], stack=False)
]),
dict(
type='Collect',
keys=['img', 'proposals', 'gt_bboxes', 'gt_labels'],
meta_keys=['scores', 'entity_ids'])
]
# The testing is w/o. any cropping / flipping
val_pipeline = [
dict(
type='SampleAVAFrames', clip_len=4, frame_interval=16, test_mode=True),
dict(type='RawFrameDecode'),
dict(type='Resize', scale=(-1, 256)),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals']),
dict(type='ToDataContainer', fields=[dict(key='proposals', stack=False)]),
dict(
type='Collect',
keys=['img', 'proposals'],
meta_keys=['scores', 'img_shape'],
nested=True)
]
data = dict(
videos_per_gpu=16,
workers_per_gpu=2,
val_dataloader=dict(videos_per_gpu=1),
test_dataloader=dict(videos_per_gpu=1),
train=dict(
type=dataset_type,
ann_file=ann_file_train,
exclude_file=exclude_file_train,
pipeline=train_pipeline,
label_file=label_file,
proposal_file=proposal_file_train,
person_det_score_thr=0.9,
data_prefix=data_root),
val=dict(
type=dataset_type,
ann_file=ann_file_val,
exclude_file=exclude_file_val,
pipeline=val_pipeline,
label_file=label_file,
proposal_file=proposal_file_val,
person_det_score_thr=0.9,
data_prefix=data_root))
data['test'] = data['val']
optimizer = dict(type='SGD', lr=0.2, momentum=0.9, weight_decay=0.00001)
# this lr is used for 8 gpus
optimizer_config = dict(grad_clip=dict(max_norm=40, norm_type=2))
# learning policy
lr_config = dict(
policy='step',
step=[10, 15],
warmup='linear',
warmup_by_epoch=True,
warmup_iters=5,
warmup_ratio=0.1)
total_epochs = 20
checkpoint_config = dict(interval=1)
workflow = [('train', 1)]
evaluation = dict(interval=1, save_best='mAP@0.5IOU')
log_config = dict(
interval=20, hooks=[
dict(type='TextLoggerHook'),
])
dist_params = dict(backend='nccl')
log_level = 'INFO'
work_dir = ('./work_dirs/ava/'
'slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb')
load_from = ('https://download.openmmlab.com/mmaction/recognition/slowonly/'
'slowonly_r50_4x16x1_256e_kinetics400_rgb/'
'slowonly_r50_4x16x1_256e_kinetics400_rgb_20200704-a69556c6.pth')
resume_from = None
find_unused_parameters = False
configs/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom_classes.py 0 → 100644
View file @ 5b3e36dc
# custom classes of ava dataset
# Here we choose classes with AP in range [0.1, 0.3)
# AP is calculated by original ckpt, which is trained by all 80 classes
custom_classes = [3, 6, 10, 27, 29, 38, 41, 48, 51, 53, 54, 59, 61, 64, 70, 72]
num_classes = len(custom_classes) + 1
# model setting
model = dict(
type='FastRCNN',
backbone=dict(
type='ResNet3dSlowOnly',
depth=50,
pretrained=None,
pretrained2d=False,
lateral=False,
num_stages=4,
conv1_kernel=(1, 7, 7),
conv1_stride_t=1,
pool1_stride_t=1,
spatial_strides=(1, 2, 2, 1)),
roi_head=dict(
type='AVARoIHead',
bbox_roi_extractor=dict(
type='SingleRoIExtractor3D',
roi_layer_type='RoIAlign',
output_size=8,
with_temporal_pool=True),
bbox_head=dict(
type='BBoxHeadAVA',
in_channels=2048,
num_classes=num_classes,
multilabel=True,
topk=(3, 5),
dropout_ratio=0.5)),
train_cfg=dict(
rcnn=dict(
assigner=dict(
type='MaxIoUAssignerAVA',
pos_iou_thr=0.9,
neg_iou_thr=0.9,
min_pos_iou=0.9),
sampler=dict(
type='RandomSampler',
num=32,
pos_fraction=1,
neg_pos_ub=-1,
add_gt_as_proposals=True),
pos_weight=1.0,
debug=False)),
test_cfg=dict(rcnn=dict(action_thr=0.002)))
dataset_type = 'AVADataset'
data_root = 'data/ava/rawframes'
anno_root = 'data/ava/annotations'
ann_file_train = f'{anno_root}/ava_train_v2.1.csv'
ann_file_val = f'{anno_root}/ava_val_v2.1.csv'
exclude_file_train = f'{anno_root}/ava_train_excluded_timestamps_v2.1.csv'
exclude_file_val = f'{anno_root}/ava_val_excluded_timestamps_v2.1.csv'
label_file = f'{anno_root}/ava_action_list_v2.1.pbtxt'
proposal_file_train = (f'{anno_root}/ava_dense_proposals_train.FAIR.'
'recall_93.9.pkl')
proposal_file_val = f'{anno_root}/ava_dense_proposals_val.FAIR.recall_93.9.pkl'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
train_pipeline = [
dict(type='SampleAVAFrames', clip_len=4, frame_interval=16),
dict(type='RawFrameDecode'),
dict(type='RandomRescale', scale_range=(256, 320)),
dict(type='RandomCrop', size=256),
dict(type='Flip', flip_ratio=0.5),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals', 'gt_bboxes', 'gt_labels']),
dict(
type='ToDataContainer',
fields=[
dict(key=['proposals', 'gt_bboxes', 'gt_labels'], stack=False)
]),
dict(
type='Collect',
keys=['img', 'proposals', 'gt_bboxes', 'gt_labels'],
meta_keys=['scores', 'entity_ids'])
]
# The testing is w/o. any cropping / flipping
val_pipeline = [
dict(
type='SampleAVAFrames', clip_len=4, frame_interval=16, test_mode=True),
dict(type='RawFrameDecode'),
dict(type='Resize', scale=(-1, 256)),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals']),
dict(type='ToDataContainer', fields=[dict(key='proposals', stack=False)]),
dict(
type='Collect',
keys=['img', 'proposals'],
meta_keys=['scores', 'img_shape'],
nested=True)
]
data = dict(
videos_per_gpu=16,
workers_per_gpu=2,
val_dataloader=dict(videos_per_gpu=1),
test_dataloader=dict(videos_per_gpu=1),
train=dict(
type=dataset_type,
ann_file=ann_file_train,
exclude_file=exclude_file_train,
pipeline=train_pipeline,
label_file=label_file,
proposal_file=proposal_file_train,
person_det_score_thr=0.9,
num_classes=num_classes,
custom_classes=custom_classes,
data_prefix=data_root),
val=dict(
type=dataset_type,
ann_file=ann_file_val,
exclude_file=exclude_file_val,
pipeline=val_pipeline,
label_file=label_file,
proposal_file=proposal_file_val,
person_det_score_thr=0.9,
num_classes=num_classes,
custom_classes=custom_classes,
data_prefix=data_root))
data['test'] = data['val']
optimizer = dict(type='SGD', lr=0.2, momentum=0.9, weight_decay=0.00001)
# this lr is used for 8 gpus
optimizer_config = dict(grad_clip=dict(max_norm=40, norm_type=2))
# learning policy
lr_config = dict(
policy='step',
step=[10, 15],
warmup='linear',
warmup_by_epoch=True,
warmup_iters=5,
warmup_ratio=0.1)
total_epochs = 20
checkpoint_config = dict(interval=1)
workflow = [('train', 1)]
evaluation = dict(interval=1, save_best='mAP@0.5IOU')
log_config = dict(
interval=20, hooks=[
dict(type='TextLoggerHook'),
])
dist_params = dict(backend='nccl')
log_level = 'INFO'
work_dir = ('./work_dirs/ava/'
'slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_custom')
load_from = ('https://download.openmmlab.com/mmaction/recognition/slowonly/'
'slowonly_r50_4x16x1_256e_kinetics400_rgb/'
'slowonly_r50_4x16x1_256e_kinetics400_rgb_20200704-a69556c6.pth')
resume_from = None
find_unused_parameters = False
configs/detection/ava/slowonly_nl_kinetics_pretrained_r50_4x16x1_10e_ava_rgb.py 0 → 100644
View file @ 5b3e36dc
_base_ = ['../_base_/models/slowonly_r50_nl.py']
dataset_type = 'AVADataset'
data_root = 'data/ava/rawframes'
anno_root = 'data/ava/annotations'
ann_file_train = f'{anno_root}/ava_train_v2.1.csv'
ann_file_val = f'{anno_root}/ava_val_v2.1.csv'
exclude_file_train = f'{anno_root}/ava_train_excluded_timestamps_v2.1.csv'
exclude_file_val = f'{anno_root}/ava_val_excluded_timestamps_v2.1.csv'
label_file = f'{anno_root}/ava_action_list_v2.1_for_activitynet_2018.pbtxt'
proposal_file_train = (f'{anno_root}/ava_dense_proposals_train.FAIR.'
'recall_93.9.pkl')
proposal_file_val = f'{anno_root}/ava_dense_proposals_val.FAIR.recall_93.9.pkl'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
train_pipeline = [
dict(type='SampleAVAFrames', clip_len=4, frame_interval=16),
dict(type='RawFrameDecode'),
dict(type='RandomRescale', scale_range=(256, 320)),
dict(type='RandomCrop', size=256),
dict(type='Flip', flip_ratio=0.5),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals', 'gt_bboxes', 'gt_labels']),
dict(
type='ToDataContainer',
fields=[
dict(key=['proposals', 'gt_bboxes', 'gt_labels'], stack=False)
]),
dict(
type='Collect',
keys=['img', 'proposals', 'gt_bboxes', 'gt_labels'],
meta_keys=['scores', 'entity_ids'])
]
# The testing is w/o. any cropping / flipping
val_pipeline = [
dict(
type='SampleAVAFrames', clip_len=4, frame_interval=16, test_mode=True),
dict(type='RawFrameDecode'),
dict(type='Resize', scale=(-1, 256)),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals']),
dict(type='ToDataContainer', fields=[dict(key='proposals', stack=False)]),
dict(
type='Collect',
keys=['img', 'proposals'],
meta_keys=['scores', 'img_shape'],
nested=True)
]
data = dict(
videos_per_gpu=12,
workers_per_gpu=2,
# During testing, each video may have different shape
val_dataloader=dict(videos_per_gpu=1),
test_dataloader=dict(videos_per_gpu=1),
train=dict(
type=dataset_type,
ann_file=ann_file_train,
exclude_file=exclude_file_train,
pipeline=train_pipeline,
label_file=label_file,
proposal_file=proposal_file_train,
person_det_score_thr=0.9,
data_prefix=data_root),
val=dict(
type=dataset_type,
ann_file=ann_file_val,
exclude_file=exclude_file_val,
pipeline=val_pipeline,
label_file=label_file,
proposal_file=proposal_file_val,
person_det_score_thr=0.9,
data_prefix=data_root))
data['test'] = data['val']
optimizer = dict(
type='SGD', lr=0.3, momentum=0.9, weight_decay=1e-06, nesterov=True)
# this lr is used for 8 gpus
optimizer_config = dict(grad_clip=dict(max_norm=40, norm_type=2))
# learning policy
lr_config = dict(
policy='step',
step=[4, 6, 8],
warmup='linear',
warmup_iters=800,
warmup_ratio=0.01)
total_epochs = 10
checkpoint_config = dict(interval=1)
workflow = [('train', 1)]
evaluation = dict(interval=1, save_best='mAP@0.5IOU')
log_config = dict(
interval=20, hooks=[
dict(type='TextLoggerHook'),
])
dist_params = dict(backend='nccl')
log_level = 'INFO'
work_dir = ('./work_dirs/ava/'
'slowonly_nl_kinetics_pretrained_r50_4x16x1_10e_ava_rgb')
load_from = (
'https://download.openmmlab.com/mmaction/recognition/slowonly/'
'slowonly_nl_embedded_gaussian_r50_4x16x1_150e_kinetics400_rgb/'
'slowonly_nl_embedded_gaussian_r50_4x16x1_150e_kinetics400_rgb_20210308-0d6e5a69.pth' # noqa: E501
)
resume_from = None
find_unused_parameters = False
configs/detection/ava/slowonly_nl_kinetics_pretrained_r50_8x8x1_10e_ava_rgb.py 0 → 100644
View file @ 5b3e36dc
_base_ = ['../_base_/models/slowonly_r50_nl.py']
dataset_type = 'AVADataset'
data_root = 'data/ava/rawframes'
anno_root = 'data/ava/annotations'
ann_file_train = f'{anno_root}/ava_train_v2.1.csv'
ann_file_val = f'{anno_root}/ava_val_v2.1.csv'
exclude_file_train = f'{anno_root}/ava_train_excluded_timestamps_v2.1.csv'
exclude_file_val = f'{anno_root}/ava_val_excluded_timestamps_v2.1.csv'
label_file = f'{anno_root}/ava_action_list_v2.1_for_activitynet_2018.pbtxt'
proposal_file_train = (f'{anno_root}/ava_dense_proposals_train.FAIR.'
'recall_93.9.pkl')
proposal_file_val = f'{anno_root}/ava_dense_proposals_val.FAIR.recall_93.9.pkl'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
train_pipeline = [
dict(type='SampleAVAFrames', clip_len=8, frame_interval=8),
dict(type='RawFrameDecode'),
dict(type='RandomRescale', scale_range=(256, 320)),
dict(type='RandomCrop', size=256),
dict(type='Flip', flip_ratio=0.5),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals', 'gt_bboxes', 'gt_labels']),
dict(
type='ToDataContainer',
fields=[
dict(key=['proposals', 'gt_bboxes', 'gt_labels'], stack=False)
]),
dict(
type='Collect',
keys=['img', 'proposals', 'gt_bboxes', 'gt_labels'],
meta_keys=['scores', 'entity_ids'])
]
# The testing is w/o. any cropping / flipping
val_pipeline = [
dict(type='SampleAVAFrames', clip_len=8, frame_interval=8, test_mode=True),
dict(type='RawFrameDecode'),
dict(type='Resize', scale=(-1, 256)),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals']),
dict(type='ToDataContainer', fields=[dict(key='proposals', stack=False)]),
dict(
type='Collect',
keys=['img', 'proposals'],
meta_keys=['scores', 'img_shape'],
nested=True)
]
data = dict(
videos_per_gpu=6,
workers_per_gpu=2,
# During testing, each video may have different shape
val_dataloader=dict(videos_per_gpu=1),
test_dataloader=dict(videos_per_gpu=1),
train=dict(
type=dataset_type,
ann_file=ann_file_train,
exclude_file=exclude_file_train,
pipeline=train_pipeline,
label_file=label_file,
proposal_file=proposal_file_train,
person_det_score_thr=0.9,
data_prefix=data_root),
val=dict(
type=dataset_type,
ann_file=ann_file_val,
exclude_file=exclude_file_val,
pipeline=val_pipeline,
label_file=label_file,
proposal_file=proposal_file_val,
person_det_score_thr=0.9,
data_prefix=data_root))
data['test'] = data['val']
optimizer = dict(
type='SGD', lr=0.15, momentum=0.9, weight_decay=1e-06, nesterov=True)
# this lr is used for 8x2 gpus
optimizer_config = dict(grad_clip=dict(max_norm=40, norm_type=2))
# learning policy
lr_config = dict(
policy='step',
step=[4, 6, 8],
warmup='linear',
warmup_iters=1600,
warmup_ratio=0.01)
total_epochs = 10
checkpoint_config = dict(interval=1)
workflow = [('train', 1)]
evaluation = dict(interval=1, save_best='mAP@0.5IOU')
log_config = dict(
interval=20, hooks=[
dict(type='TextLoggerHook'),
])
dist_params = dict(backend='nccl')
log_level = 'INFO'
work_dir = ('./work_dirs/ava/'
'slowonly_nl_kinetics_pretrained_r50_8x8x1_10e_ava_rgb')
load_from = (
'https://download.openmmlab.com/mmaction/recognition/slowonly/'
'slowonly_nl_embedded_gaussian_r50_8x8x1_150e_kinetics400_rgb/'
'slowonly_nl_embedded_gaussian_r50_8x8x1_150e_kinetics400_rgb_20210308-e8dd9e82.pth' # noqa: E501
)
resume_from = None
find_unused_parameters = False
configs/detection/ava/slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb.py 0 → 100644
View file @ 5b3e36dc
# model setting
model = dict(
type='FastRCNN',
backbone=dict(
type='ResNet3dSlowOnly',
depth=101,
pretrained=None,
pretrained2d=False,
lateral=False,
num_stages=4,
conv1_kernel=(1, 7, 7),
conv1_stride_t=1,
pool1_stride_t=1,
spatial_strides=(1, 2, 2, 1)),
roi_head=dict(
type='AVARoIHead',
bbox_roi_extractor=dict(
type='SingleRoIExtractor3D',
roi_layer_type='RoIAlign',
output_size=8,
with_temporal_pool=True),
bbox_head=dict(
type='BBoxHeadAVA',
in_channels=2048,
num_classes=81,
multilabel=True,
dropout_ratio=0.5)),
train_cfg=dict(
rcnn=dict(
assigner=dict(
type='MaxIoUAssignerAVA',
pos_iou_thr=0.9,
neg_iou_thr=0.9,
min_pos_iou=0.9),
sampler=dict(
type='RandomSampler',
num=32,
pos_fraction=1,
neg_pos_ub=-1,
add_gt_as_proposals=True),
pos_weight=1.0,
debug=False)),
test_cfg=dict(rcnn=dict(action_thr=0.002)))
dataset_type = 'AVADataset'
data_root = 'data/ava/rawframes'
anno_root = 'data/ava/annotations'
ann_file_train = f'{anno_root}/ava_train_v2.1.csv'
ann_file_val = f'{anno_root}/ava_val_v2.1.csv'
exclude_file_train = f'{anno_root}/ava_train_excluded_timestamps_v2.1.csv'
exclude_file_val = f'{anno_root}/ava_val_excluded_timestamps_v2.1.csv'
label_file = f'{anno_root}/ava_action_list_v2.1_for_activitynet_2018.pbtxt'
proposal_file_train = (f'{anno_root}/ava_dense_proposals_train.FAIR.'
'recall_93.9.pkl')
proposal_file_val = f'{anno_root}/ava_dense_proposals_val.FAIR.recall_93.9.pkl'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
train_pipeline = [
dict(type='SampleAVAFrames', clip_len=8, frame_interval=8),
dict(type='RawFrameDecode'),
dict(type='RandomRescale', scale_range=(256, 320)),
dict(type='RandomCrop', size=256),
dict(type='Flip', flip_ratio=0.5),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals', 'gt_bboxes', 'gt_labels']),
dict(
type='ToDataContainer',
fields=[
dict(key=['proposals', 'gt_bboxes', 'gt_labels'], stack=False)
]),
dict(
type='Collect',
keys=['img', 'proposals', 'gt_bboxes', 'gt_labels'],
meta_keys=['scores', 'entity_ids'])
]
# The testing is w/o. any cropping / flipping
val_pipeline = [
dict(type='SampleAVAFrames', clip_len=8, frame_interval=8, test_mode=True),
dict(type='RawFrameDecode'),
dict(type='Resize', scale=(-1, 256)),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals']),
dict(type='ToDataContainer', fields=[dict(key='proposals', stack=False)]),
dict(
type='Collect',
keys=['img', 'proposals'],
meta_keys=['scores', 'img_shape'],
nested=True)
]
data = dict(
videos_per_gpu=6,
workers_per_gpu=2,
# During testing, each video may have different shape
val_dataloader=dict(videos_per_gpu=1),
test_dataloader=dict(videos_per_gpu=1),
train=dict(
type=dataset_type,
ann_file=ann_file_train,
exclude_file=exclude_file_train,
pipeline=train_pipeline,
label_file=label_file,
proposal_file=proposal_file_train,
person_det_score_thr=0.9,
data_prefix=data_root),
val=dict(
type=dataset_type,
ann_file=ann_file_val,
exclude_file=exclude_file_val,
pipeline=val_pipeline,
label_file=label_file,
proposal_file=proposal_file_val,
person_det_score_thr=0.9,
data_prefix=data_root))
data['test'] = data['val']
optimizer = dict(type='SGD', lr=0.075, momentum=0.9, weight_decay=0.00001)
# this lr is used for 8 gpus
optimizer_config = dict(grad_clip=dict(max_norm=40, norm_type=2))
# learning policy
lr_config = dict(
policy='step',
step=[10, 15],
warmup='linear',
warmup_by_epoch=True,
warmup_iters=5,
warmup_ratio=0.1)
total_epochs = 20
checkpoint_config = dict(interval=1)
workflow = [('train', 1)]
evaluation = dict(interval=1, save_best='mAP@0.5IOU')
log_config = dict(
interval=20, hooks=[
dict(type='TextLoggerHook'),
])
dist_params = dict(backend='nccl')
log_level = 'INFO'
work_dir = ('./work_dirs/ava/'
'slowonly_omnisource_pretrained_r101_8x8x1_20e_ava_rgb')
load_from = ('https://download.openmmlab.com/mmaction/recognition/slowonly/'
'omni/'
'slowonly_r101_omni_8x8x1_kinetics400_rgb_20200926-b5dbb701.pth')
resume_from = None
find_unused_parameters = False
configs/detection/ava/slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb.py 0 → 100644
View file @ 5b3e36dc
# model setting
model = dict(
type='FastRCNN',
backbone=dict(
type='ResNet3dSlowOnly',
depth=50,
pretrained=None,
pretrained2d=False,
lateral=False,
num_stages=4,
conv1_kernel=(1, 7, 7),
conv1_stride_t=1,
pool1_stride_t=1,
spatial_strides=(1, 2, 2, 1)),
roi_head=dict(
type='AVARoIHead',
bbox_roi_extractor=dict(
type='SingleRoIExtractor3D',
roi_layer_type='RoIAlign',
output_size=8,
with_temporal_pool=True),
bbox_head=dict(
type='BBoxHeadAVA',
in_channels=2048,
num_classes=81,
multilabel=True,
dropout_ratio=0.5)),
train_cfg=dict(
rcnn=dict(
assigner=dict(
type='MaxIoUAssignerAVA',
pos_iou_thr=0.9,
neg_iou_thr=0.9,
min_pos_iou=0.9),
sampler=dict(
type='RandomSampler',
num=32,
pos_fraction=1,
neg_pos_ub=-1,
add_gt_as_proposals=True),
pos_weight=1.0,
debug=False)),
test_cfg=dict(rcnn=dict(action_thr=0.002)))
dataset_type = 'AVADataset'
data_root = 'data/ava/rawframes'
anno_root = 'data/ava/annotations'
ann_file_train = f'{anno_root}/ava_train_v2.1.csv'
ann_file_val = f'{anno_root}/ava_val_v2.1.csv'
exclude_file_train = f'{anno_root}/ava_train_excluded_timestamps_v2.1.csv'
exclude_file_val = f'{anno_root}/ava_val_excluded_timestamps_v2.1.csv'
label_file = f'{anno_root}/ava_action_list_v2.1_for_activitynet_2018.pbtxt'
proposal_file_train = (f'{anno_root}/ava_dense_proposals_train.FAIR.'
'recall_93.9.pkl')
proposal_file_val = f'{anno_root}/ava_dense_proposals_val.FAIR.recall_93.9.pkl'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_bgr=False)
train_pipeline = [
dict(type='SampleAVAFrames', clip_len=4, frame_interval=16),
dict(type='RawFrameDecode'),
dict(type='RandomRescale', scale_range=(256, 320)),
dict(type='RandomCrop', size=256),
dict(type='Flip', flip_ratio=0.5),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals', 'gt_bboxes', 'gt_labels']),
dict(
type='ToDataContainer',
fields=[
dict(key=['proposals', 'gt_bboxes', 'gt_labels'], stack=False)
]),
dict(
type='Collect',
keys=['img', 'proposals', 'gt_bboxes', 'gt_labels'],
meta_keys=['scores', 'entity_ids'])
]
# The testing is w/o. any cropping / flipping
val_pipeline = [
dict(
type='SampleAVAFrames', clip_len=4, frame_interval=16, test_mode=True),
dict(type='RawFrameDecode'),
dict(type='Resize', scale=(-1, 256)),
dict(type='Normalize', **img_norm_cfg),
dict(type='FormatShape', input_format='NCTHW', collapse=True),
# Rename is needed to use mmdet detectors
dict(type='Rename', mapping=dict(imgs='img')),
dict(type='ToTensor', keys=['img', 'proposals']),
dict(type='ToDataContainer', fields=[dict(key='proposals', stack=False)]),
dict(
type='Collect',
keys=['img', 'proposals'],
meta_keys=['scores', 'img_shape'],
nested=True)
]
data = dict(
videos_per_gpu=16,
workers_per_gpu=2,
# During testing, each video may have different shape
val_dataloader=dict(videos_per_gpu=1),
test_dataloader=dict(videos_per_gpu=1),
train=dict(
type=dataset_type,
ann_file=ann_file_train,
exclude_file=exclude_file_train,
pipeline=train_pipeline,
label_file=label_file,
proposal_file=proposal_file_train,
person_det_score_thr=0.9,
data_prefix=data_root),
val=dict(
type=dataset_type,
ann_file=ann_file_val,
exclude_file=exclude_file_val,
pipeline=val_pipeline,
label_file=label_file,
proposal_file=proposal_file_val,
person_det_score_thr=0.9,
data_prefix=data_root))
data['test'] = data['val']
optimizer = dict(type='SGD', lr=0.2, momentum=0.9, weight_decay=0.00001)
# this lr is used for 8 gpus
optimizer_config = dict(grad_clip=dict(max_norm=40, norm_type=2))
# learning policy
lr_config = dict(
policy='step',
step=[10, 15],
warmup='linear',
warmup_by_epoch=True,
warmup_iters=5,
warmup_ratio=0.1)
total_epochs = 20
checkpoint_config = dict(interval=1)
workflow = [('train', 1)]
evaluation = dict(interval=1, save_best='mAP@0.5IOU')
log_config = dict(
interval=20, hooks=[
dict(type='TextLoggerHook'),
])
dist_params = dict(backend='nccl')
log_level = 'INFO'
work_dir = ('./work_dirs/ava/'
'slowonly_omnisource_pretrained_r50_4x16x1_20e_ava_rgb')
load_from = ('https://download.openmmlab.com/mmaction/recognition/slowonly/'
'omni/'
'slowonly_r50_omni_4x16x1_kinetics400_rgb_20200926-51b1f7ea.pth')
resume_from = None
find_unused_parameters = False
configs/detection/lfb/README.md 0 → 100644
View file @ 5b3e36dc
# LFB
[Long-term feature banks for detailed video understanding](https://openaccess.thecvf.com/content_CVPR_2019/html/Wu_Long-Term_Feature_Banks_for_Detailed_Video_Understanding_CVPR_2019_paper.html)
<!-- [ALGORITHM] -->
## Abstract
<!-- [ABSTRACT] -->
To understand the world, we humans constantly need to relate the present to the past, and put events in context. In this paper, we enable existing video models to do the same. We propose a long-term feature bank---supportive information extracted over the entire span of a video---to augment state-of-the-art video models that otherwise would only view short clips of 2-5 seconds. Our experiments demonstrate that augmenting 3D convolutional networks with a long-term feature bank yields state-of-the-art results on three challenging video datasets: AVA, EPIC-Kitchens, and Charades.
<!-- [IMAGE] -->
<div align=center>
<img src="https://user-images.githubusercontent.com/34324155/143016220-21d90fb3-fd9f-499c-820f-f6c421bda7aa.png" width="800"/>
</div>
## Results and Models
### AVA2.1
| Model | Modality | Pretrained | Backbone | Input | gpus | Resolution | mAP | log | json | ckpt |
| :-----------------------------------------------------------------------------------------------------------------------------------------------------: | :------: | :----------: | :--------------------------------------------------------------------------------------------------: | :---: | :--: | :------------: | :---: | :------------------------------------------------------------------------------------------------------------------------------------------: | :------------------------------------------------------------------------------------------------------------------------------------------------: | :---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
| [lfb_nl_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb.py](/configs/detection/lfb/lfb_nl_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb.py) | RGB | Kinetics-400 | [slowonly_r50_4x16x1](/configs/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py) | 4x16 | 8 | short-side 256 | 24.11 | [log](https://download.openmmlab.com/mmaction/detection/lfb/lfb_nl_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/20210224_125052.log) | [json](https://download.openmmlab.com/mmaction/detection/lfb/lfb_nl_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/20210224_125052.log.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/lfb/lfb_nl_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/lfb_nl_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb_20210224-2ae136d9.pth) |
| [lfb_avg_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb.py](/configs/detection/lfb/lfb_avg_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb.py) | RGB | Kinetics-400 | [slowonly_r50_4x16x1](/configs/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py) | 4x16 | 8 | short-side 256 | 20.17 | [log](https://download.openmmlab.com/mmaction/detection/lfb/lfb_avg_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/20210301_124812.log) | [json](https://download.openmmlab.com/mmaction/detection/lfb/lfb_avg_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/20210301_124812.log.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/lfb/lfb_avg_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/lfb_avg_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb_20210301-19c330b7.pth) |
| [lfb_max_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb.py](/configs/detection/lfb/lfb_max_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb.py) | RGB | Kinetics-400 | [slowonly_r50_4x16x1](/configs/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py) | 4x16 | 8 | short-side 256 | 22.15 | [log](https://download.openmmlab.com/mmaction/detection/lfb/lfb_max_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/20210301_124812.log) | [json](https://download.openmmlab.com/mmaction/detection/lfb/lfb_max_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/20210301_124812.log.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/lfb/lfb_max_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/lfb_max_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb_20210301-37efcd15.pth) |
:::{note}
1. The **gpus** indicates the number of gpu we used to get the checkpoint.
According to the [Linear Scaling Rule](https://arxiv.org/abs/1706.02677), you may set the learning rate proportional to the batch size if you use different GPUs or videos per GPU,
e.g., lr=0.01 for 4 GPUs x 2 video/gpu and lr=0.08 for 16 GPUs x 4 video/gpu.
2. We use `slowonly_r50_4x16x1` instead of `I3D-R50-NL` in the original paper as the backbone of LFB, but we have achieved the similar improvement: (ours: 20.1 -> 24.11 vs. author: 22.1 -> 25.8).
3. Because the long-term features are randomly sampled in testing, the test accuracy may have some differences.
4. Before train or test lfb, you need to infer feature bank with the [lfb_slowonly_r50_ava_infer.py](/configs/detection/lfb/lfb_slowonly_r50_ava_infer.py). For more details on infer feature bank, you can refer to [Train](#Train) part.
5. You can also dowonload long-term feature bank from [AVA_train_val_float32_lfb](https://download.openmmlab.com/mmaction/detection/lfb/AVA_train_val_float32_lfb.rar) or [AVA_train_val_float16_lfb](https://download.openmmlab.com/mmaction/detection/lfb/AVA_train_val_float16_lfb.rar), and then put them on `lfb_prefix_path`.
6. The ROIHead now supports single-label classification (i.e. the network outputs at most
one-label per actor). This can be done by (a) setting multilabel=False during training and
the test_cfg.rcnn.action_thr for testing.
:::
## Train
### a. Infer long-term feature bank for training
Before train or test lfb, you need to infer long-term feature bank first.
Specifically, run the test on the training, validation, testing dataset with the config file [lfb_slowonly_r50_ava_infer](/configs/detection/lfb/lfb_slowonly_r50_ava_infer.py) (The config file will only infer the feature bank of training dataset and you need set `dataset_mode = 'val'` to infer the feature bank of validation dataset in the config file.), and the shared head [LFBInferHead](/mmaction/models/heads/lfb_infer_head.py) will generate the feature bank.
A long-term feature bank file of AVA training and validation datasets with float32 precision occupies 3.3 GB. If store the features with float16 precision, the feature bank occupies 1.65 GB.
You can use the following command to infer feature bank of AVA training and validation dataset and the feature bank will be stored in `lfb_prefix_path/lfb_train.pkl` and `lfb_prefix_path/lfb_val.pkl`.
```shell
# set `dataset_mode = 'train'` in lfb_slowonly_r50_ava_infer.py
python tools/test.py configs/detection/lfb/lfb_slowonly_r50_ava_infer.py \
checkpoints/YOUR_BASELINE_CHECKPOINT.pth --eval mAP
# set `dataset_mode = 'val'` in lfb_slowonly_r50_ava_infer.py
python tools/test.py configs/detection/lfb/lfb_slowonly_r50_ava_infer.py \
checkpoints/YOUR_BASELINE_CHECKPOINT.pth --eval mAP
```
We use [slowonly_r50_4x16x1 checkpoint](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217-40061d5f.pth) from [slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb](/configs/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py) to infer feature bank.
### b. Train LFB
You can use the following command to train a model.
```shell
python tools/train.py ${CONFIG_FILE} [optional arguments]
```
Example: train LFB model on AVA with half-precision long-term feature bank.
```shell
python tools/train.py configs/detection/lfb/lfb_nl_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb.py \
--validate --seed 0 --deterministic
```
For more details and optional arguments infos, you can refer to **Training setting** part in [getting_started](/docs/en/getting_started.md#training-setting).
## Test
### a. Infer long-term feature bank for testing
Before train or test lfb, you also need to infer long-term feature bank first. If you have generated the feature bank file, you can skip it.
The step is the same with **Infer long-term feature bank for training** part in [Train](#Train).
### b. Test LFB
You can use the following command to test a model.
```shell
python tools/test.py ${CONFIG_FILE} ${CHECKPOINT_FILE} [optional arguments]
```
Example: test LFB model on AVA with half-precision long-term feature bank and dump the result to a csv file.
```shell
python tools/test.py configs/detection/lfb/lfb_nl_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb.py \
checkpoints/SOME_CHECKPOINT.pth --eval mAP --out results.csv
```
For more details, you can refer to **Test a dataset** part in [getting_started](/docs/en/getting_started.md#test-a-dataset).
## Citation
<!-- [DATASET] -->
```BibTeX
@inproceedings{gu2018ava,
title={Ava: A video dataset of spatio-temporally localized atomic visual actions},
author={Gu, Chunhui and Sun, Chen and Ross, David A and Vondrick, Carl and Pantofaru, Caroline and Li, Yeqing and Vijayanarasimhan, Sudheendra and Toderici, George and Ricco, Susanna and Sukthankar, Rahul and others},
booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
pages={6047--6056},
year={2018}
}
```
```BibTeX
@inproceedings{wu2019long,
title={Long-term feature banks for detailed video understanding},
author={Wu, Chao-Yuan and Feichtenhofer, Christoph and Fan, Haoqi and He, Kaiming and Krahenbuhl, Philipp and Girshick, Ross},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
pages={284--293},
year={2019}
}
```
configs/detection/lfb/README_zh-CN.md 0 → 100644
View file @ 5b3e36dc
# LFB
## 简介
<!-- [ALGORITHM] -->
```BibTeX
@inproceedings{wu2019long,
title={Long-term feature banks for detailed video understanding},
author={Wu, Chao-Yuan and Feichtenhofer, Christoph and Fan, Haoqi and He, Kaiming and Krahenbuhl, Philipp and Girshick, Ross},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
pages={284--293},
year={2019}
}
```
## 模型库
### AVA2.1
| 配置文件 | 模态 | 预训练 | 主干网络 | 输入 | GPU 数量 | 分辨率 | 平均精度 | log | json | ckpt |
| :-----------------------------------------------------------------------------------------------------------------------------------------------------: | :--: | :----------: | :--------------------------------------------------------------------------------------------------: | :--: | :------: | :------: | :------: | :------------------------------------------------------------------------------------------------------------------------------------------: | :------------------------------------------------------------------------------------------------------------------------------------------------: | :---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
| [lfb_nl_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb.py](/configs/detection/lfb/lfb_nl_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb.py) | RGB | Kinetics-400 | [slowonly_r50_4x16x1](/configs/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py) | 4x16 | 8 | 短边 256 | 24.11 | [log](https://download.openmmlab.com/mmaction/detection/lfb/lfb_nl_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/20210224_125052.log) | [json](https://download.openmmlab.com/mmaction/detection/lfb/lfb_nl_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/20210224_125052.log.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/lfb/lfb_nl_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/lfb_nl_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb_20210224-2ae136d9.pth) |
| [lfb_avg_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb.py](/configs/detection/lfb/lfb_avg_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb.py) | RGB | Kinetics-400 | [slowonly_r50_4x16x1](/configs/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py) | 4x16 | 8 | 短边 256 | 20.17 | [log](https://download.openmmlab.com/mmaction/detection/lfb/lfb_avg_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/20210301_124812.log) | [json](https://download.openmmlab.com/mmaction/detection/lfb/lfb_avg_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/20210301_124812.log.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/lfb/lfb_avg_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/lfb_avg_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb_20210301-19c330b7.pth) |
| [lfb_max_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb.py](/configs/detection/lfb/lfb_max_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb.py) | RGB | Kinetics-400 | [slowonly_r50_4x16x1](/configs/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py) | 4x16 | 8 | 短边 256 | 22.15 | [log](https://download.openmmlab.com/mmaction/detection/lfb/lfb_max_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/20210301_124812.log) | [json](https://download.openmmlab.com/mmaction/detection/lfb/lfb_max_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/20210301_124812.log.json) | [ckpt](https://download.openmmlab.com/mmaction/detection/lfb/lfb_max_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/lfb_max_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb_20210301-37efcd15.pth) |
- 注:
1. 这里的 **GPU 数量** 指的是得到模型权重文件对应的 GPU 个数。默认地，MMAction2 所提供的配置文件对应使用 8 块 GPU 进行训练的情况。
依据 [线性缩放规则](https://arxiv.org/abs/1706.02677)，当用户使用不同数量的 GPU 或者每块 GPU 处理不同视频个数时，需要根据批大小等比例地调节学习率。
如，lr=0.01 对应 4 GPUs x 2 video/gpu，以及 lr=0.08 对应 16 GPUs x 4 video/gpu。
2. 本 LFB 模型暂没有使用原论文中的 `I3D-R50-NL` 作为主干网络，而是用 `slowonly_r50_4x16x1` 替代，但取得了同样的提升效果：（本模型：20.1 -> 24.11 而原论文模型：22.1 -> 25.8）。
3. 因为测试时，长时特征是被随机采样的，所以测试精度可能有一些偏差。
4. 在训练或测试 LFB 之前，用户需要使用配置文件特征库 [lfb_slowonly_r50_ava_infer.py](/configs/detection/lfb/lfb_slowonly_r50_ava_infer.py) 来推导长时特征库。有关推导长时特征库的更多细节，请参照[训练部分](#%E8%AE%AD%E7%BB%83)。
5. 用户也可以直接从 [AVA_train_val_float32_lfb](https://download.openmmlab.com/mmaction/detection/lfb/AVA_train_val_float32_lfb.rar) 或者 [AVA_train_val_float16_lfb](https://download.openmmlab.com/mmaction/detection/lfb/AVA_train_val_float16_lfb.rar) 下载 float32 或 float16 的长时特征库，并把它们放在 `lfb_prefix_path` 上。
## 训练
### a. 为训练 LFB 推导长时特征库
在训练或测试 LFB 之前，用户首先需要推导长时特征库。
具体来说，使用配置文件 [lfb_slowonly_r50_ava_infer](/configs/detection/lfb/lfb_slowonly_r50_ava_infer.py)，在训练集、验证集、测试集上都运行一次模型测试。
配置文件的默认设置是推导训练集的长时特征库，用户需要将 `dataset_mode` 设置成 `'val'` 来推导验证集的长时特征库，在推导过程中。共享头 [LFBInferHead](/mmaction/models/heads/lfb_infer_head.py) 会生成长时特征库。
AVA 训练集和验证集的 float32 精度的长时特征库文件大约占 3.3 GB。如果以半精度来存储长时特征，文件大约占 1.65 GB。
用户可以使用以下命令来推导 AVA 训练集和验证集的长时特征库，而特征库会被存储为 `lfb_prefix_path/lfb_train.pkl` 和 `lfb_prefix_path/lfb_val.pkl`。
```shell
# 在 lfb_slowonly_r50_ava_infer.py 中 设置 `dataset_mode = 'train'`
python tools/test.py configs/detection/lfb/lfb_slowonly_r50_ava_infer.py \
checkpoints/YOUR_BASELINE_CHECKPOINT.pth --eval mAP
# 在 lfb_slowonly_r50_ava_infer.py 中 设置 `dataset_mode = 'val'`
python tools/test.py configs/detection/lfb/lfb_slowonly_r50_ava_infer.py \
checkpoints/YOUR_BASELINE_CHECKPOINT.pth --eval mAP
```
MMAction2 使用来自配置文件 [slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb](/configs/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb.py) 的模型权重文件 [slowonly_r50_4x16x1 checkpoint](https://download.openmmlab.com/mmaction/detection/ava/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb/slowonly_kinetics_pretrained_r50_4x16x1_20e_ava_rgb_20201217-40061d5f.pth)作为推导长时特征库的 LFB 模型的主干网络的预训练模型。
### b. 训练 LFB
用户可以使用以下指令进行模型训练。
```shell
python tools/train.py ${CONFIG_FILE} [optional arguments]
```
例如：使用半精度的长时特征库在 AVA 数据集上训练 LFB 模型。
```shell
python tools/train.py configs/detection/lfb/lfb_nl_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb.py \
--validate --seed 0 --deterministic
```
更多训练细节，可参考 [基础教程](/docs/zh_cn/getting_started.md#训练配置) 中的 **训练配置** 部分。
## 测试
### a. 为测试 LFB 推导长时特征库
在训练或测试 LFB 之前，用户首先需要推导长时特征库。如果用户之前已经生成了特征库文件，可以跳过这一步。
这一步做法与[训练部分](#Train)中的 **为训练 LFB 推导长时特征库** 相同。
### b. 测试 LFB
用户可以使用以下指令进行模型测试。
```shell
python tools/test.py ${CONFIG_FILE} ${CHECKPOINT_FILE} [optional arguments]
```
例如：使用半精度的长时特征库在 AVA 数据集上测试 LFB 模型，并将结果导出为一个 json 文件。
```shell
python tools/test.py configs/detection/lfb/lfb_nl_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb.py \
checkpoints/SOME_CHECKPOINT.pth --eval mAP --out results.csv
```
更多测试细节，可参考 [基础教程](/docs/zh_cn/getting_started.md#测试某个数据集) 中的 **测试某个数据集** 部分。
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment