"vscode:/vscode.git/clone" did not exist on "b0f41f60bdc9c1c76a2e28fecc64aa8da277e354"
Unverified Commit d7067e44 authored by Wenwei Zhang's avatar Wenwei Zhang Committed by GitHub
Browse files

Bump version to v1.1.0rc2

Bump to v1.1.0rc2
parents 28fe73d2 fb0e57e5
......@@ -91,12 +91,8 @@ test_pipeline = [
dict(type='RandomFlip3D', sync_2d=False),
dict(
type='PointsRangeFilter', point_cloud_range=point_cloud_range),
dict(
type='DefaultFormatBundle3D',
class_names=class_names,
with_label=False),
dict(type='Collect3D', keys=['points'])
])
]),
dict(type='Pack3DDetInputs', keys=['points'])
]
data = dict(
......@@ -109,23 +105,25 @@ data = dict(
### CenterPoint
| Backbone | Voxel type (voxel size) | Dcn | Circular nms | Mem (GB) | Inf time (fps) | mAP | NDS | Download |
| :----------------------------------------------------------------------------------------: | :---------------------: | :-: | :----------: | :------: | :------------: | :---: | :---: | :--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
| [SECFPN](./centerpoint_voxel01_second_secfpn_head-circlenms_8xb4-cyclic-20e_nus-3d.py) | voxel (0.1) | ✗ | ✓ | 4.9 | | 56.19 | 64.43 | [model](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_01voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus/centerpoint_01voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus_20210815_085857-9ba7f3a5.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_01voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus/centerpoint_01voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus_20210815_085857.log.json) |
| above w/o circle nms | voxel (0.1) | ✗ | ✗ | | | 56.56 | 64.46 | |
| [SECFPN](./centerpoint_voxel01_second_secfpn_head-dcn-circlenms_8xb4-cyclic-20e_nus-3d.py) | voxel (0.1) | ✓ | ✓ | 5.2 | | 56.34 | 64.81 | [model](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_01voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus/centerpoint_01voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus_20210814_060754-c9d535d2.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_01voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus/centerpoint_01voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus_20210814_060754.log.json) |
| above w/o circle nms | voxel (0.1) | ✓ | ✗ | | | 56.60 | 64.90 | |
| [SECFPN](./centerpoint_voxel0075_second_secfpn_head-circlenms_8xb4-cyclic-20e_nus-3d.py) | voxel (0.075) | ✗ | ✓ | 7.8 | | 57.34 | 65.23 | [model](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_0075voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus/centerpoint_0075voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus_20210814_113418-76ae0cf0.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_0075voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus/centerpoint_0075voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus_20210814_113418.log.json) |
| :---------------------------------------------------------------------------------: | :---------------------: | :-: | :----------: | :------: | :------------: | :---: | :---: | :---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
| [SECFPN](./centerpoint_01voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus.py) | voxel (0.1) | ✗ | ✓ | 5.2 | | 56.11 | 64.61 | [model](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_01voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus/centerpoint_01voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus_20220810_030004-9061688e.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_01voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus/centerpoint_01voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus_20220810_030004.log) |
| above w/o circle nms | voxel (0.1) | ✗ | ✗ | | | x | x | |
| [SECFPN](./centerpoint_01voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus.py) | voxel (0.1) | ✓ | ✓ | 5.5 | | 56.10 | 64.69 | [model](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_01voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus/centerpoint_01voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus_20220810_052355-a6928835.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_01voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus/centerpoint_01voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus_20220810_052355.log) |
| above w/o circle nms | voxel (0.1) | ✓ | ✗ | | | x | x | |
| [SECFPN](./centerpoint_0075voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus.py) | voxel (0.075) | ✗ | ✓ | 8.2 | | 56.54 | 65.17 | [model](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_0075voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus/centerpoint_0075voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus_20220810_011659-04cb3a3b.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_0075voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus/centerpoint_0075voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus_20220810_011659.log) |
| above w/o circle nms | voxel (0.075) | ✗ | ✗ | | | 57.63 | 65.39 | |
| [SECFPN](./centerpoint_voxel0075-second-secfpn-head-dcn=circlenms_8xb4-cyclic-20e_nus.py) | voxel (0.075) | ✓ | ✓ | 8.5 | | 57.27 | 65.58 | [model](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_0075voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus/centerpoint_0075voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus_20210827_161135-1782af3e.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_0075voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus/centerpoint_0075voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus_20210827_161135.log.json) |
| [SECFPN](./centerpoint_0075voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus.py) | voxel (0.075) | ✓ | ✓ | 8.7 | | 56.92 | 65.27 | [model](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_0075voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus/centerpoint_0075voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus_20220810_025930-657f67e0.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_0075voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus/centerpoint_0075voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus_20220810_025930.log) |
| above w/o circle nms | voxel (0.075) | ✓ | ✗ | | | 57.43 | 65.63 | |
| above w/ double flip | voxel (0.075) | ✓ | ✗ | | | 59.73 | 67.39 | |
| above w/ scale tta | voxel (0.075) | ✓ | ✗ | | | 60.43 | 67.65 | |
| above w/ circle nms w/o scale tta | voxel (0.075) | ✓ | ✗ | | | 59.52 | 67.24 | |
| [SECFPN](./centerpoint_pillar02_second_secfpn_head-circlenms_8xb4-cyclic-20e_nus-3d.py) | pillar (0.2) | ✗ | ✓ | 4.4 | | 49.07 | 59.66 | [model](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_02pillar_second_secfpn_circlenms_4x8_cyclic_20e_nus/centerpoint_02pillar_second_secfpn_circlenms_4x8_cyclic_20e_nus_20210816_064624-0f3299c0.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_02pillar_second_secfpn_circlenms_4x8_cyclic_20e_nus/centerpoint_02pillar_second_secfpn_circlenms_4x8_cyclic_20e_nus_20210816_064624.log.json) |
| [SECFPN](./centerpoint_02pillar_second_secfpn_circlenms_4x8_cyclic_20e_nus.py) | pillar (0.2) | ✗ | ✓ | 4.6 | | 48.70 | 59.62 | [model](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_02pillar_second_secfpn_circlenms_4x8_cyclic_20e_nus/centerpoint_02pillar_second_secfpn_circlenms_4x8_cyclic_20e_nus_20220811_031844-191a3822.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_02pillar_second_secfpn_circlenms_4x8_cyclic_20e_nus/centerpoint_02pillar_second_secfpn_circlenms_4x8_cyclic_20e_nus_20220811_031844.log) |
| above w/o circle nms | pillar (0.2) | ✗ | ✗ | | | 49.12 | 59.66 | |
| [SECFPN](./centerpoint_pillar02_second_secfpn_head-dcn_8xb4-cyclic-20e_nus-3d.py) | pillar (0.2) | ✓ | ✗ | 4.6 | | 48.8 | 59.67 | [model](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_02pillar_second_secfpn_dcn_4x8_cyclic_20e_nus/centerpoint_02pillar_second_secfpn_dcn_4x8_cyclic_20e_nus_20210815_202702-f03ab9e4.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_02pillar_second_secfpn_dcn_4x8_cyclic_20e_nus/centerpoint_02pillar_second_secfpn_dcn_4x8_cyclic_20e_nus_20210815_202702.log.json) |
| [SECFPN](./centerpoint_02pillar_second_secfpn_dcn_4x8_cyclic_20e_nus.py) | pillar (0.2) | ✓ | ✗ | 4.9 | | 48.38 | 59.79 | [model](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_02pillar_second_secfpn_dcn_4x8_cyclic_20e_nus/centerpoint_02pillar_second_secfpn_dcn_4x8_cyclic_20e_nus_20220811_045458-808e69ad.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_02pillar_second_secfpn_dcn_4x8_cyclic_20e_nus/centerpoint_02pillar_second_secfpn_dcn_4x8_cyclic_20e_nus_20220811_045458.log) |
| above w/ circle nms | pillar (0.2) | ✓ | ✓ | | | 48.79 | 59.65 | |
**Note:** The model performance after coordinate refactor is slightly different (+/- 0.5 - 1 mAP/NDS) from the performance before coordinate refactor in v0.x branch. We are exploring the reason behind. |
## Citation
```latex
......
......@@ -7,6 +7,9 @@ _base_ = [
# If point cloud range is changed, the models should also change their point
# cloud range accordingly
point_cloud_range = [-51.2, -51.2, -5.0, 51.2, 51.2, 3.0]
# Using calibration info convert the Lidar-coordinate point cloud range to the
# ego-coordinate point cloud range could bring a little promotion in nuScenes.
# point_cloud_range = [-51.2, -52, -5.0, 51.2, 50.4, 3.0]
# For nuScenes we usually do 10-class detection
class_names = [
'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',
......@@ -126,7 +129,7 @@ train_dataloader = dict(
data_root=data_root,
ann_file='nuscenes_infos_train.pkl',
pipeline=train_pipeline,
metainfo=dict(CLASSES=class_names),
metainfo=dict(classes=class_names),
test_mode=False,
data_prefix=data_prefix,
use_valid_flag=True,
......@@ -134,8 +137,8 @@ train_dataloader = dict(
# and box_type_3d='Depth' in sunrgbd and scannet dataset.
box_type_3d='LiDAR')))
test_dataloader = dict(
dataset=dict(pipeline=test_pipeline, metainfo=dict(CLASSES=class_names)))
dataset=dict(pipeline=test_pipeline, metainfo=dict(classes=class_names)))
val_dataloader = dict(
dataset=dict(pipeline=test_pipeline, metainfo=dict(CLASSES=class_names)))
dataset=dict(pipeline=test_pipeline, metainfo=dict(classes=class_names)))
train_cfg = dict(val_interval=20)
......@@ -4,6 +4,9 @@ _base_ = ['./centerpoint_voxel01_second_secfpn_8xb4-cyclic-20e_nus-3d.py']
# cloud range accordingly
voxel_size = [0.075, 0.075, 0.2]
point_cloud_range = [-54, -54, -5.0, 54, 54, 3.0]
# Using calibration info convert the Lidar-coordinate point cloud range to the
# ego-coordinate point cloud range could bring a little promotion in nuScenes.
# point_cloud_range = [-54, -54.8, -5.0, 54, 53.2, 3.0]
# For nuScenes we usually do 10-class detection
class_names = [
'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',
......@@ -89,7 +92,6 @@ train_pipeline = [
dict(type='ObjectRangeFilter', point_cloud_range=point_cloud_range),
dict(type='ObjectNameFilter', classes=class_names),
dict(type='PointShuffle'),
dict(type='DefaultFormatBundle3D', class_names=class_names),
dict(
type='Pack3DDetInputs',
keys=['points', 'gt_bboxes_3d', 'gt_labels_3d'])
......@@ -122,8 +124,8 @@ test_pipeline = [
train_dataloader = dict(
dataset=dict(
dataset=dict(
pipeline=train_pipeline, metainfo=dict(CLASSES=class_names))))
pipeline=train_pipeline, metainfo=dict(classes=class_names))))
test_dataloader = dict(
dataset=dict(pipeline=test_pipeline, metainfo=dict(CLASSES=class_names)))
dataset=dict(pipeline=test_pipeline, metainfo=dict(classes=class_names)))
val_dataloader = dict(
dataset=dict(pipeline=test_pipeline, metainfo=dict(CLASSES=class_names)))
dataset=dict(pipeline=test_pipeline, metainfo=dict(classes=class_names)))
......@@ -2,6 +2,9 @@ _base_ = './centerpoint_voxel0075_second_secfpn_' \
'head-dcn-circlenms_8xb4_cyclic-20e_nus-3d.py'
point_cloud_range = [-54, -54, -5.0, 54, 54, 3.0]
# Using calibration info convert the Lidar-coordinate point cloud range to the
# ego-coordinate point cloud range could bring a little promotion in nuScenes.
# point_cloud_range = [-54, -54.8, -5.0, 54, 53.2, 3.0]
file_client_args = dict(backend='disk')
class_names = [
'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',
......
......@@ -2,6 +2,9 @@ _base_ = './centerpoint_voxel0075_second_secfpn' \
'_head-dcn_8xb4-cyclic-20e_nus-3d.py'
point_cloud_range = [-54, -54, -5.0, 54, 54, 3.0]
# Using calibration info convert the Lidar-coordinate point cloud range to the
# ego-coordinate point cloud range could bring a little promotion in nuScenes.
# point_cloud_range = [-54, -54.8, -5.0, 54, 53.2, 3.0]
file_client_args = dict(backend='disk')
class_names = [
'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',
......
......@@ -4,6 +4,9 @@ _base_ = './centerpoint_voxel0075_second_secfpn' \
model = dict(test_cfg=dict(pts=dict(use_rotate_nms=True, max_num=500)))
point_cloud_range = [-54, -54, -5.0, 54, 54, 3.0]
# Using calibration info convert the Lidar-coordinate point cloud range to the
# ego-coordinate point cloud range could bring a little promotion in nuScenes.
# point_cloud_range = [-54, -54.8, -5.0, 54, 53.2, 3.0]
file_client_args = dict(backend='disk')
class_names = [
'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',
......
......@@ -7,6 +7,9 @@ _base_ = [
# If point cloud range is changed, the models should also change their point
# cloud range accordingly
point_cloud_range = [-51.2, -51.2, -5.0, 51.2, 51.2, 3.0]
# Using calibration info convert the Lidar-coordinate point cloud range to the
# ego-coordinate point cloud range could bring a little promotion in nuScenes.
# point_cloud_range = [-51.2, -52, -5.0, 51.2, 50.4, 3.0]
# For nuScenes we usually do 10-class detection
class_names = [
'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',
......@@ -127,7 +130,7 @@ train_dataloader = dict(
data_root=data_root,
ann_file='nuscenes_infos_train.pkl',
pipeline=train_pipeline,
metainfo=dict(CLASSES=class_names),
metainfo=dict(classes=class_names),
test_mode=False,
data_prefix=data_prefix,
use_valid_flag=True,
......@@ -135,8 +138,8 @@ train_dataloader = dict(
# and box_type_3d='Depth' in sunrgbd and scannet dataset.
box_type_3d='LiDAR')))
test_dataloader = dict(
dataset=dict(pipeline=test_pipeline, metainfo=dict(CLASSES=class_names)))
dataset=dict(pipeline=test_pipeline, metainfo=dict(classes=class_names)))
val_dataloader = dict(
dataset=dict(pipeline=test_pipeline, metainfo=dict(CLASSES=class_names)))
dataset=dict(pipeline=test_pipeline, metainfo=dict(classes=class_names)))
train_cfg = dict(val_interval=20)
......@@ -16,80 +16,80 @@ Collections:
Version: v0.6.0
Models:
- Name: centerpoint_01voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus
- Name: centerpoint_voxel01_second_secfpn_head-circlenms_8xb4-cyclic-20e_nus-3d
In Collection: CenterPoint
Config: configs/centerpoint/centerpoint_voxel01_second_secfpn_head-circlenms_8xb4-cyclic-20e_nus-3d.py
Metadata:
Training Memory (GB): 4.9
metadata:
Training Memory (GB): 5.2
Results:
- Task: 3D Object Detection
Dataset: nuScenes
Metrics:
mAP: 56.19
NDS: 64.43
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/centerpoint/centerpoint_01voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus/centerpoint_01voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus_20201001_135205-5db91e00.pth
mAP: 56.11
NDS: 64.61
Weights: https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_01voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus/centerpoint_01voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus_20220810_030004-9061688e.pth
- Name: centerpoint_01voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus
- Name: centerpoint_voxel01_second_secfpn_head-dcn-circlenms_8xb4-cyclic-20e_nus-3d
In Collection: CenterPoint
Config: configs/centerpoint/centerpoint_voxel01_second_secfpn_head-dcn-circlenms_8xb4-cyclic-20e_nus-3d.py
Metadata:
Training Memory (GB): 5.2
Training Memory (GB): 5.5
Results:
- Task: 3D Object Detection
Dataset: nuScenes
Metrics:
mAP: 56.34
NDS: 64.81
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/centerpoint/centerpoint_01voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus/centerpoint_01voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus_20201004_075317-26d8176c.pth
mAP: 56.10
NDS: 64.69
Weights: https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_01voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus/centerpoint_01voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus_20220810_052355-a6928835.pth
- Name: centerpoint_0075voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus
- Name: centerpoint_voxel0075_second_secfpn_head-circlenms_8xb4-cyclic-20e_nus-3d
In Collection: CenterPoint
Config: configs/centerpoint/centerpoint_voxel0075_second_secfpn_head-circlenms_8xb4-cyclic-20e_nus-3d.py
Metadata:
Training Memory (GB): 7.8
Training Memory (GB): 8.2
Results:
- Task: 3D Object Detection
Dataset: nuScenes
Metrics:
mAP: 57.34
NDS: 65.23
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/centerpoint/centerpoint_0075voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus/centerpoint_0075voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus_20200925_230905-358fbe3b.pth
mAP: 56.54
NDS: 65.17
Weights: https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_0075voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus/centerpoint_0075voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus_20220810_011659-04cb3a3b.pth
- Name: centerpoint_0075voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus
- Name: centerpoint_voxel0075_second_secfpn_head-dcn-circlenms_8xb4-cyclic-20e_nus-3d
In Collection: CenterPoint
Config: configs/centerpoint/centerpoint_voxel0075_second_secfpn_head-dcn-circlenms_8xb4-cyclic-20e_nus-3d.py
Metadata:
Training Memory (GB): 8.5
Training Memory (GB): 8.7
Results:
- Task: 3D Object Detection
Dataset: nuScenes
Metrics:
mAP: 57.27
NDS: 65.58
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/centerpoint/centerpoint_0075voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus/centerpoint_0075voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus_20200930_201619-67c8496f.pth
mAP: 56.92
NDS: 65.27
Weights: https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_0075voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus/centerpoint_0075voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus_20220810_025930-657f67e0.pth
- Name: centerpoint_02pillar_second_secfpn_circlenms_4x8_cyclic_20e_nus
- Name: centerpoint_pillar02_second_secfpn_head-circlenms_8xb4-cyclic-20e_nus-3d
In Collection: CenterPoint
Config: configs/centerpoint/centerpoint_pillar02_second_secfpn_head-circlenms_8xb4-cyclic-20e_nus-3d.py
Metadata:
Training Memory (GB): 4.4
Training Memory (GB): 4.6
Results:
- Task: 3D Object Detection
Dataset: nuScenes
Metrics:
mAP: 49.07
NDS: 59.66
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/centerpoint/centerpoint_01voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus/centerpoint_02pillar_second_secfpn_circlenms_4x8_cyclic_20e_nus_20201004_170716-a134a233.pth
mAP: 48.70
NDS: 59.62
Weights: https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_02pillar_second_secfpn_circlenms_4x8_cyclic_20e_nus/centerpoint_02pillar_second_secfpn_circlenms_4x8_cyclic_20e_nus_20220811_031844-191a3822.pth
- Name: centerpoint_02pillar_second_secfpn_dcn_4x8_cyclic_20e_nus
- Name: centerpoint_pillar02_second_secfpn_head-dcn_8xb4-cyclic-20e_nus-3d
In Collection: CenterPoint
Config: configs/centerpoint/centerpoint_pillar02_second_secfpn_head-dcn_8xb4-cyclic-20e_nus-3d.py
Metadata:
Training Memory (GB): 4.6
Training Memory (GB): 4.9
Results:
- Task: 3D Object Detection
Dataset: nuScenes
Metrics:
mAP: 48.8
NDS: 59.67
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/centerpoint/centerpoint_02pillar_second_secfpn_dcn_4x8_cyclic_20e_nus/centerpoint_02pillar_second_secfpn_dcn_4x8_cyclic_20e_nus_20200930_103722-3bb135f2.pth
mAP: 48.38
NDS: 59.79
Weights: https://download.openmmlab.com/mmdetection3d/v1.0.0_models/centerpoint/centerpoint_02pillar_second_secfpn_dcn_4x8_cyclic_20e_nus/centerpoint_02pillar_second_secfpn_dcn_4x8_cyclic_20e_nus_20220811_045458-808e69ad.pth
......@@ -23,14 +23,14 @@ We implement DGCNN and provide the results and checkpoints on S3DIS dataset.
### S3DIS
| Method | Split | Lr schd | Mem (GB) | Inf time (fps) | mIoU (Val set) | Download |
| :---------------------------------------------: | :----: | :---------: | :------: | :------------: | :------------: | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
| [DGCNN](./dgcnn_4xb32-cosine-100e_s3dis-seg.py) | Area_1 | cosine 100e | 13.1 | | 68.33 | [model](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area1/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210731_000734-39658f14.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area1/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210731_000734.log.json) |
| [DGCNN](./dgcnn_4xb32-cosine-100e_s3dis-seg.py) | Area_2 | cosine 100e | 13.1 | | 40.68 | [model](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area2/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210731_144648-aea9ecb6.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area2/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210731_144648.log.json) |
| [DGCNN](./dgcnn_4xb32-cosine-100e_s3dis-seg.py) | Area_3 | cosine 100e | 13.1 | | 69.38 | [model](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area3/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210801_154629-2ff50ee0.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area3/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210801_154629.log.json) |
| [DGCNN](./dgcnn_4xb32-cosine-100e_s3dis-seg.py) | Area_4 | cosine 100e | 13.1 | | 50.07 | [model](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area4/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210802_073551-dffab9cd.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area4/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210802_073551.log.json) |
| [DGCNN](./dgcnn_4xb32-cosine-100e_s3dis-seg.py) | Area_5 | cosine 100e | 13.1 | | 50.59 | [model](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area5/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210730_235824-f277e0c5.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area5/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210730_235824.log.json) |
| [DGCNN](./dgcnn_4xb32-cosine-100e_s3dis-seg.py) | Area_6 | cosine 100e | 13.1 | | 77.94 | [model](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area6/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210802_154317-e3511b32.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area6/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210802_154317.log.json) |
| [DGCNN](./dgcnn_4xb32-cosine-100e_s3dis-seg.py) | 6-fold | | | | 59.43 | |
| :--------------------------------------------------------: | :----: | :---------: | :------: | :------------: | :------------: | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
| [DGCNN](./dgcnn_4xb32-cosine-100e_s3dis-seg_test-area1.py) | Area_1 | cosine 100e | 13.1 | | 68.33 | [model](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area1/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210731_000734-39658f14.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area1/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210731_000734.log.json) |
| [DGCNN](./dgcnn_4xb32-cosine-100e_s3dis-seg_test-area2.py) | Area_2 | cosine 100e | 13.1 | | 40.68 | [model](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area2/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210731_144648-aea9ecb6.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area2/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210731_144648.log.json) |
| [DGCNN](./dgcnn_4xb32-cosine-100e_s3dis-seg_test-area3.py) | Area_3 | cosine 100e | 13.1 | | 69.38 | [model](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area3/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210801_154629-2ff50ee0.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area3/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210801_154629.log.json) |
| [DGCNN](./dgcnn_4xb32-cosine-100e_s3dis-seg_test-area4.py) | Area_4 | cosine 100e | 13.1 | | 50.07 | [model](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area4/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210802_073551-dffab9cd.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area4/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210802_073551.log.json) |
| [DGCNN](./dgcnn_4xb32-cosine-100e_s3dis-seg_test-area5.py) | Area_5 | cosine 100e | 13.1 | | 50.59 | [model](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area5/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210730_235824-f277e0c5.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area5/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210730_235824.log.json) |
| [DGCNN](./dgcnn_4xb32-cosine-100e_s3dis-seg_test-area6.py) | Area_6 | cosine 100e | 13.1 | | 77.94 | [model](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area6/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210802_154317-e3511b32.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area6/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210802_154317.log.json) |
| DGCNN | 6-fold | | | | 59.43 | |
**Notes:**
......
_base_ = './dgcnn_4xb32-cosine-100e_s3dis-seg_test-area5.py'
# data settings
train_area = [2, 3, 4, 5, 6]
test_area = 1
train_dataloader = dict(
batch_size=32,
dataset=dict(
ann_files=[f's3dis_infos_Area_{i}.pkl' for i in train_area],
scene_idxs=[
f'seg_info/Area_{i}_resampled_scene_idxs.npy' for i in train_area
]))
test_dataloader = dict(
dataset=dict(
ann_files=f's3dis_infos_Area_{test_area}.pkl',
scene_idxs=f'seg_info/Area_{test_area}_resampled_scene_idxs.npy'))
val_dataloader = test_dataloader
_base_ = './dgcnn_4xb32-cosine-100e_s3dis-seg_test-area5.py'
# data settings
train_area = [1, 3, 4, 5, 6]
test_area = 2
train_dataloader = dict(
batch_size=32,
dataset=dict(
ann_files=[f's3dis_infos_Area_{i}.pkl' for i in train_area],
scene_idxs=[
f'seg_info/Area_{i}_resampled_scene_idxs.npy' for i in train_area
]))
test_dataloader = dict(
dataset=dict(
ann_files=f's3dis_infos_Area_{test_area}.pkl',
scene_idxs=f'seg_info/Area_{test_area}_resampled_scene_idxs.npy'))
val_dataloader = test_dataloader
_base_ = './dgcnn_4xb32-cosine-100e_s3dis-seg_test-area5.py'
# data settings
train_area = [1, 2, 4, 5, 6]
test_area = 3
train_dataloader = dict(
batch_size=32,
dataset=dict(
ann_files=[f's3dis_infos_Area_{i}.pkl' for i in train_area],
scene_idxs=[
f'seg_info/Area_{i}_resampled_scene_idxs.npy' for i in train_area
]))
test_dataloader = dict(
dataset=dict(
ann_files=f's3dis_infos_Area_{test_area}.pkl',
scene_idxs=f'seg_info/Area_{test_area}_resampled_scene_idxs.npy'))
val_dataloader = test_dataloader
_base_ = './dgcnn_4xb32-cosine-100e_s3dis-seg_test-area5.py'
# data settings
train_area = [1, 2, 3, 5, 6]
test_area = 4
train_dataloader = dict(
batch_size=32,
dataset=dict(
ann_files=[f's3dis_infos_Area_{i}.pkl' for i in train_area],
scene_idxs=[
f'seg_info/Area_{i}_resampled_scene_idxs.npy' for i in train_area
]))
test_dataloader = dict(
dataset=dict(
ann_files=f's3dis_infos_Area_{test_area}.pkl',
scene_idxs=f'seg_info/Area_{test_area}_resampled_scene_idxs.npy'))
val_dataloader = test_dataloader
......@@ -16,6 +16,6 @@ model = dict(
use_normalized_coord=True,
batch_size=24))
default_hooks = dict(checkpoint=dict(type='CheckpointHook', interval=2), )
default_hooks = dict(checkpoint=dict(type='CheckpointHook', interval=2))
train_dataloader = dict(batch_size=32)
val_cfg = dict(interval=2)
train_cfg = dict(val_interval=2)
_base_ = './dgcnn_4xb32-cosine-100e_s3dis-seg_test-area5.py'
# data settings
train_area = [1, 2, 3, 4, 5]
test_area = 6
train_dataloader = dict(
batch_size=32,
dataset=dict(
ann_files=[f's3dis_infos_Area_{i}.pkl' for i in train_area],
scene_idxs=[
f'seg_info/Area_{i}_resampled_scene_idxs.npy' for i in train_area
]))
test_dataloader = dict(
dataset=dict(
ann_files=f's3dis_infos_Area_{test_area}.pkl',
scene_idxs=f'seg_info/Area_{test_area}_resampled_scene_idxs.npy'))
val_dataloader = test_dataloader
......@@ -10,15 +10,80 @@ Collections:
README: configs/dgcnn/README.md
Models:
- Name: dgcnn_4xb32-cosine-100e_s3dis-seg.py
- Name: dgcnn_4xb32-cosine-100e_s3dis-seg_test-area1.py
In Collection: DGCNN
Config: configs/dgcnn/dgcnn_4xb32-cosine-100e_s3dis-seg.py
Config: configs/dgcnn/dgcnn_4xb32-cosine-100e_s3dis-seg_test-area1.py
Metadata:
Training Data: S3DIS
Training Memory (GB): 13.3
Results:
- Task: 3D Semantic Segmentation
Dataset: S3DIS
Dataset: S3DIS Area1
Metrics:
mIoU: 68.33
Weights: https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area1/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210731_000734-39658f14.pth
- Name: dgcnn_4xb32-cosine-100e_s3dis-seg_test-area2.py
In Collection: DGCNN
Config: configs/dgcnn/dgcnn_4xb32-cosine-100e_s3dis-seg_test-area2.py
Metadata:
Training Data: S3DIS
Training Memory (GB): 13.3
Results:
- Task: 3D Semantic Segmentation
Dataset: S3DIS Area2
Metrics:
mIoU: 40.68
Weights: https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area2/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210731_144648-aea9ecb6.pth
- Name: dgcnn_4xb32-cosine-100e_s3dis-seg_test-area3.py
In Collection: DGCNN
Config: configs/dgcnn/dgcnn_4xb32-cosine-100e_s3dis-seg_test-area3.py
Metadata:
Training Data: S3DIS
Training Memory (GB): 13.3
Results:
- Task: 3D Semantic Segmentation
Dataset: S3DIS Area3
Metrics:
mIoU: 69.38
Weights: https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area3/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210801_154629-2ff50ee0.pth
- Name: dgcnn_4xb32-cosine-100e_s3dis-seg_test-area4.py
In Collection: DGCNN
Config: configs/dgcnn/dgcnn_4xb32-cosine-100e_s3dis-seg_test-area4.py
Metadata:
Training Data: S3DIS
Training Memory (GB): 13.3
Results:
- Task: 3D Semantic Segmentation
Dataset: S3DIS Area4
Metrics:
mIoU: 50.07
Weights: https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area4/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210802_073551-dffab9cd.pth
- Name: dgcnn_4xb32-cosine-100e_s3dis-seg_test-area5.py
In Collection: DGCNN
Config: configs/dgcnn/dgcnn_4xb32-cosine-100e_s3dis-seg_test-area5.py
Metadata:
Training Data: S3DIS
Training Memory (GB): 13.3
Results:
- Task: 3D Semantic Segmentation
Dataset: S3DIS Area5
Metrics:
mIoU: 50.59
Weights: https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area5/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210730_235824-f277e0c5.pth
- Name: dgcnn_4xb32-cosine-100e_s3dis-seg_test-area6.py
In Collection: DGCNN
Config: configs/dgcnn/dgcnn_4xb32-cosine-100e_s3dis-seg_test-area6.py
Metadata:
Training Data: S3DIS
Training Memory (GB): 13.3
Results:
- Task: 3D Semantic Segmentation
Dataset: S3DIS Area6
Metrics:
mIoU: 77.94
Weights: https://download.openmmlab.com/mmdetection3d/v0.17.0_models/dgcnn/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class/area6/dgcnn_32x4_cosine_100e_s3dis_seg-3d-13class_20210802_154317-e3511b32.pth
# FCAF3D: Fully Convolutional Anchor-Free 3D Object Detection
> [FCAF3D: Fully Convolutional Anchor-Free 3D Object Detection](https://arxiv.org/abs/2112.00322)
<!-- [ALGORITHM] -->
## Abstract
Recently, promising applications in robotics and augmented reality have attracted considerable attention to 3D object detection from point clouds. In this paper, we present FCAF3D --- a first-in-class fully convolutional anchor-free indoor 3D object detection method. It is a simple yet effective method that uses a voxel representation of a point cloud and processes voxels with sparse convolutions. FCAF3D can handle large-scale scenes with minimal runtime through a single fully convolutional feed-forward pass. Existing 3D object detection methods make prior assumptions on the geometry of objects, and we argue that it limits their generalization ability. To eliminate prior assumptions, we propose a novel parametrization of oriented bounding boxes that allows obtaining better results in a purely data-driven way. The proposed method achieves state-of-the-art 3D object detection results in terms of mAP@0.5 on ScanNet V2 (+4.5), SUN RGB-D (+3.5), and S3DIS (+20.5) datasets.
<div align="center">
<img src="https://user-images.githubusercontent.com/6030962/182842796-98c10576-d39c-4c2b-a15a-a04c9870919c.png" width="800"/>
</div>
## Introduction
We implement FCAF3D and provide the result and checkpoints on the ScanNet and SUN RGB-D dataset.
## Results and models
### ScanNet
| Backbone | Mem (GB) | Inf time (fps) | AP@0.25 | AP@0.5 | Download |
| :------------------------------------------------: | :------: | :------------: | :----------: | :----------: | :-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
| [MinkResNet34](./fcaf3d_8x2_scannet-3d-18class.py) | 10.5 | 8.0 | 69.7(70.7\*) | 55.2(56.0\*) | [model](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/fcaf3d/fcaf3d_8x2_scannet-3d-18class/fcaf3d_8x2_scannet-3d-18class_20220805_084956.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/fcaf3d/fcaf3d_8x2_scannet-3d-18class/fcaf3d_8x2_scannet-3d-18class_20220805_084956.log.json) |
### SUN RGB-D
| Backbone | Mem (GB) | Inf time (fps) | AP@0.25 | AP@0.5 | Download |
| :------------------------------------------------: | :------: | :------------: | :----------: | :----------: | :-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
| [MinkResNet34](./fcaf3d_8x2_sunrgbd-3d-10class.py) | 6.3 | 15.6 | 63.8(63.8\*) | 47.3(48.2\*) | [model](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/fcaf3d/fcaf3d_8x2_sunrgbd-3d-10class/fcaf3d_8x2_sunrgbd-3d-10class_20220805_165017.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/fcaf3d/fcaf3d_8x2_sunrgbd-3d-10class/fcaf3d_8x2_sunrgbd-3d-10class_20220805_165017.log.json) |
### S3DIS
| Backbone | Mem (GB) | Inf time (fps) | AP@0.25 | AP@0.5 | Download |
| :----------------------------------------------: | :------: | :------------: | :----------: | :----------: | :-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
| [MinkResNet34](./fcaf3d_2xb8_s3dis-3d-5class.py) | 23.5 | 4.2 | 67.4(64.9\*) | 45.7(43.8\*) | [model](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/fcaf3d/fcaf3d_8x2_s3dis-3d-5class/fcaf3d_8x2_s3dis-3d-5class_20220805_121957.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/fcaf3d/fcaf3d_8x2_s3dis-3d-5class/fcaf3d_8x2_s3dis-3d-5class_20220805_121957.log.json) |
**Note**
- We report the results across 5 train runs followed by 5 test runs. * means the results reported in the paper.
- Inference time is given for a single NVidia GeForce GTX 1080 Ti GPU. All models are trained on 2 GPUs.
## Citation
```latex
@inproceedings{rukhovich2022fcaf3d,
title={FCAF3D: Fully Convolutional Anchor-Free 3D Object Detection},
author={Danila Rukhovich, Anna Vorontsova, Anton Konushin},
booktitle={European conference on computer vision},
year={2022}
}
```
_base_ = [
'../_base_/models/fcaf3d.py', '../_base_/default_runtime.py',
'../_base_/datasets/s3dis-3d.py'
]
model = dict(bbox_head=dict(num_classes=5))
optim_wrapper = dict(
type='OptimWrapper',
optimizer=dict(type='AdamW', lr=0.001, weight_decay=0.0001),
clip_grad=dict(max_norm=10, norm_type=2))
# learning rate
param_scheduler = dict(
type='MultiStepLR',
begin=0,
end=12,
by_epoch=True,
milestones=[8, 11],
gamma=0.1)
custom_hooks = [dict(type='EmptyCacheHook', after_iter=True)]
# training schedule for 1x
train_cfg = dict(type='EpochBasedTrainLoop', max_epochs=12, val_interval=12)
val_cfg = dict(type='ValLoop')
test_cfg = dict(type='TestLoop')
_base_ = [
'../_base_/models/fcaf3d.py', '../_base_/default_runtime.py',
'../_base_/datasets/scannet-3d.py'
]
n_points = 100000
train_pipeline = [
dict(
type='LoadPointsFromFile',
coord_type='DEPTH',
shift_height=False,
use_color=True,
load_dim=6,
use_dim=[0, 1, 2, 3, 4, 5]),
dict(type='LoadAnnotations3D'),
dict(type='GlobalAlignment', rotation_axis=2),
dict(type='PointSample', num_points=n_points),
dict(
type='RandomFlip3D',
sync_2d=False,
flip_ratio_bev_horizontal=0.5,
flip_ratio_bev_vertical=0.5),
dict(
type='GlobalRotScaleTrans',
rot_range=[-0.087266, 0.087266],
scale_ratio_range=[.9, 1.1],
translation_std=[.1, .1, .1],
shift_height=False),
dict(type='NormalizePointsColor', color_mean=None),
dict(
type='Pack3DDetInputs',
keys=['points', 'gt_bboxes_3d', 'gt_labels_3d'])
]
test_pipeline = [
dict(
type='LoadPointsFromFile',
coord_type='DEPTH',
shift_height=False,
use_color=True,
load_dim=6,
use_dim=[0, 1, 2, 3, 4, 5]),
dict(type='GlobalAlignment', rotation_axis=2),
dict(
type='MultiScaleFlipAug3D',
img_scale=(1333, 800),
pts_scale_ratio=1,
flip=False,
transforms=[
dict(
type='GlobalRotScaleTrans',
rot_range=[0, 0],
scale_ratio_range=[1., 1.],
translation_std=[0, 0, 0]),
dict(
type='RandomFlip3D',
sync_2d=False,
flip_ratio_bev_horizontal=0.5,
flip_ratio_bev_vertical=0.5),
dict(type='PointSample', num_points=n_points),
dict(type='NormalizePointsColor', color_mean=None),
]),
dict(type='Pack3DDetInputs', keys=['points'])
]
train_dataloader = dict(
dataset=dict(
type='RepeatDataset',
times=10,
dataset=dict(pipeline=train_pipeline, filter_empty_gt=True)))
val_dataloader = dict(dataset=dict(pipeline=test_pipeline))
test_dataloader = val_dataloader
optim_wrapper = dict(
type='OptimWrapper',
optimizer=dict(type='AdamW', lr=0.001, weight_decay=0.0001),
clip_grad=dict(max_norm=10, norm_type=2))
# learning rate
param_scheduler = dict(
type='MultiStepLR',
begin=0,
end=12,
by_epoch=True,
milestones=[8, 11],
gamma=0.1)
custom_hooks = [dict(type='EmptyCacheHook', after_iter=True)]
# training schedule for 1x
train_cfg = dict(type='EpochBasedTrainLoop', max_epochs=12, val_interval=12)
val_cfg = dict(type='ValLoop')
test_cfg = dict(type='TestLoop')
_base_ = [
'../_base_/models/fcaf3d.py', '../_base_/default_runtime.py',
'../_base_/datasets/sunrgbd-3d.py'
]
n_points = 100000
model = dict(
bbox_head=dict(
num_classes=10,
num_reg_outs=8,
bbox_loss=dict(type='RotatedIoU3DLoss')))
train_pipeline = [
dict(
type='LoadPointsFromFile',
coord_type='DEPTH',
shift_height=False,
load_dim=6,
use_dim=[0, 1, 2, 3, 4, 5]),
dict(type='LoadAnnotations3D'),
dict(type='PointSample', num_points=n_points),
dict(type='RandomFlip3D', sync_2d=False, flip_ratio_bev_horizontal=0.5),
dict(
type='GlobalRotScaleTrans',
rot_range=[-0.523599, 0.523599],
scale_ratio_range=[0.85, 1.15],
translation_std=[.1, .1, .1],
shift_height=False),
dict(
type='Pack3DDetInputs',
keys=['points', 'gt_bboxes_3d', 'gt_labels_3d'])
]
test_pipeline = [
dict(
type='LoadPointsFromFile',
coord_type='DEPTH',
shift_height=False,
load_dim=6,
use_dim=[0, 1, 2, 3, 4, 5]),
dict(
type='MultiScaleFlipAug3D',
img_scale=(1333, 800),
pts_scale_ratio=1,
flip=False,
transforms=[
dict(
type='GlobalRotScaleTrans',
rot_range=[0, 0],
scale_ratio_range=[1., 1.],
translation_std=[0, 0, 0]),
dict(
type='RandomFlip3D',
sync_2d=False,
flip_ratio_bev_horizontal=0.5,
flip_ratio_bev_vertical=0.5),
dict(type='PointSample', num_points=n_points)
]),
dict(type='Pack3DDetInputs', keys=['points'])
]
train_dataloader = dict(
batch_size=8,
dataset=dict(
type='RepeatDataset',
times=3,
dataset=dict(pipeline=train_pipeline, filter_empty_gt=True)))
val_dataloader = dict(dataset=dict(pipeline=test_pipeline))
test_dataloader = val_dataloader
optim_wrapper = dict(
type='OptimWrapper',
optimizer=dict(type='AdamW', lr=0.001, weight_decay=0.0001),
clip_grad=dict(max_norm=10, norm_type=2))
# learning rate
param_scheduler = dict(
type='MultiStepLR',
begin=0,
end=12,
by_epoch=True,
milestones=[8, 11],
gamma=0.1)
custom_hooks = [dict(type='EmptyCacheHook', after_iter=True)]
# training schedule for 1x
train_cfg = dict(type='EpochBasedTrainLoop', max_epochs=12, val_interval=12)
val_cfg = dict(type='ValLoop')
test_cfg = dict(type='TestLoop')
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