[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.
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
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.
For more details and optional arguments infos, you can refer to **Training setting** part in [getting_started](/docs/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.
For more details, you can refer to **Test a dataset** part in [getting_started](/docs/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},
Training Json Log:https://download.openmmlab.com/mmaction/detection/lfb/lfb_nl_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/20210224_125052.log.json
Training Log:https://download.openmmlab.com/mmaction/detection/lfb/lfb_nl_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/20210224_125052.log
Training Json Log:https://download.openmmlab.com/mmaction/detection/lfb/lfb_avg_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/20210301_124812.log.json
Training Log:https://download.openmmlab.com/mmaction/detection/lfb/lfb_avg_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/20210301_124812.log
Training Json Log:https://download.openmmlab.com/mmaction/detection/lfb/lfb_max_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/20210301_124812.log.json
Training Log:https://download.openmmlab.com/mmaction/detection/lfb/lfb_max_kinetics_pretrained_slowonly_r50_4x16x1_20e_ava_rgb/20210301_124812.log
[Bmn: Boundary-matching network for temporal action proposal generation](https://openaccess.thecvf.com/content_ICCV_2019/html/Lin_BMN_Boundary-Matching_Network_for_Temporal_Action_Proposal_Generation_ICCV_2019_paper.html)
<!-- [ALGORITHM] -->
## Abstract
<!-- [ABSTRACT] -->
Temporal action proposal generation is an challenging and promising task which aims to locate temporal regions in real-world videos where action or event may occur. Current bottom-up proposal generation methods can generate proposals with precise boundary, but cannot efficiently generate adequately reliable confidence scores for retrieving proposals. To address these difficulties, we introduce the Boundary-Matching (BM) mechanism to evaluate confidence scores of densely distributed proposals, which denote a proposal as a matching pair of starting and ending boundaries and combine all densely distributed BM pairs into the BM confidence map. Based on BM mechanism, we propose an effective, efficient and end-to-end proposal generation method, named Boundary-Matching Network (BMN), which generates proposals with precise temporal boundaries as well as reliable confidence scores simultaneously. The two-branches of BMN are jointly trained in an unified framework. We conduct experiments on two challenging datasets: THUMOS-14 and ActivityNet-1.3, where BMN shows significant performance improvement with remarkable efficiency and generalizability. Further, combining with existing action classifier, BMN can achieve state-of-the-art temporal action detection performance.
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. For feature column, cuhk_mean_100 denotes the widely used cuhk activitynet feature extracted by [anet2016-cuhk](https://github.com/yjxiong/anet2016-cuhk), mmaction_video and mmaction_clip denote feature extracted by mmaction, with video-level activitynet finetuned model or clip-level activitynet finetuned model respectively.
3. We evaluate the action detection performance of BMN, using [anet_cuhk_2017](https://download.openmmlab.com/mmaction/localization/cuhk_anet17_pred.json) submission for ActivityNet2017 Untrimmed Video Classification Track to assign label for each action proposal.
:::
\*We train BMN with the [official repo](https://github.com/JJBOY/BMN-Boundary-Matching-Network), evaluate its proposal generation and action detection performance with [anet_cuhk_2017](https://download.openmmlab.com/mmaction/localization/cuhk_anet17_pred.json) for label assigning.
For more details on data preparation, you can refer to ActivityNet feature in [Data Preparation](/docs/data_preparation.md).
## Train
You can use the following command to train a model.
For more details and optional arguments infos, you can refer to **Training setting** part in [getting_started](/docs/getting_started.md#training-setting) .
## Test
You can use the following command to test a model.
You can also test the action detection performance of the model, with [anet_cuhk_2017](https://download.openmmlab.com/mmaction/localization/cuhk_anet17_pred.json) prediction file and generated proposal file (`results.json` in last command).
1. (Optional) You can use the following command to generate a formatted proposal file, which will be fed into the action classifier (Currently supports SSN and P-GCN, not including TSN, I3D etc.) to get the classification result of proposals.
For more details and optional arguments infos, you can refer to **Test a dataset** part in [getting_started](/docs/getting_started.md#test-a-dataset) .
## Citation
```BibTeX
@inproceedings{lin2019bmn,
title={Bmn: Boundary-matching network for temporal action proposal generation},
author={Lin, Tianwei and Liu, Xiao and Li, Xin and Ding, Errui and Wen, Shilei},
booktitle={Proceedings of the IEEE International Conference on Computer Vision},
pages={3889--3898},
year={2019}
}
```
<!-- [DATASET] -->
```BibTeX
@article{zhao2017cuhk,
title={Cuhk \& ethz \& siat submission to activitynet challenge 2017},
author={Zhao, Y and Zhang, B and Wu, Z and Yang, S and Zhou, L and Yan, S and Wang, L and Xiong, Y and Lin, D and Qiao, Y and others},
Training Json Log:https://download.openmmlab.com/mmaction/localization/bmn/bmn_400x100_9e_activitynet_feature/bmn_400x100_9e_activitynet_feature.log.json
Training Log:https://download.openmmlab.com/mmaction/localization/bmn/bmn_400x100_9e_activitynet_feature/bmn_400x100_9e_activitynet_feature.log
Training Json Log:https://download.openmmlab.com/mmaction/localization/bmn/bmn_400x100_2x8_9e_mmaction_video/bmn_400x100_2x8_9e_mmaction_video_20200809.json
Training Log:https://download.openmmlab.com/mmaction/localization/bmn/bmn_400x100_2x8_9e_mmaction_video/bmn_400x100_2x8_9e_mmaction_video_20200809.log
Training Json Log:https://download.openmmlab.com/mmaction/localization/bmn/bmn_400x100_2x8_9e_mmaction_clip/bmn_400x100_2x8_9e_mmaction_clip_20200809.json
Training Log:https://download.openmmlab.com/mmaction/localization/bmn/bmn_400x100_2x8_9e_mmaction_clip/bmn_400x100_2x8_9e_mmaction_clip_20200809.log
[Bsn: Boundary sensitive network for temporal action proposal generation](https://openaccess.thecvf.com/content_ECCV_2018/html/Tianwei_Lin_BSN_Boundary_Sensitive_ECCV_2018_paper.html)
<!-- [ALGORITHM] -->
## Abstract
<!-- [ABSTRACT] -->
Temporal action proposal generation is an important yet challenging problem, since temporal proposals with rich action content are indispensable for analysing real-world videos with long duration and high proportion irrelevant content. This problem requires methods not only generating proposals with precise temporal boundaries, but also retrieving proposals to cover truth action instances with high recall and high overlap using relatively fewer proposals. To address these difficulties, we introduce an effective proposal generation method, named Boundary-Sensitive Network (BSN), which adopts "local to global" fashion. Locally, BSN first locates temporal boundaries with high probabilities, then directly combines these boundaries as proposals. Globally, with Boundary-Sensitive Proposal feature, BSN retrieves proposals by evaluating the confidence of whether a proposal contains an action within its region. We conduct experiments on two challenging datasets: ActivityNet-1.3 and THUMOS14, where BSN outperforms other state-of-the-art temporal action proposal generation methods with high recall and high temporal precision. Finally, further experiments demonstrate that by combining existing action classifiers, our method significantly improves the state-of-the-art temporal action detection performance.
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. For feature column, cuhk_mean_100 denotes the widely used cuhk activitynet feature extracted by [anet2016-cuhk](https://github.com/yjxiong/anet2016-cuhk), mmaction_video and mmaction_clip denote feature extracted by mmaction, with video-level activitynet finetuned model or clip-level activitynet finetuned model respectively.
:::
For more details on data preparation, you can refer to ActivityNet feature in [Data Preparation](/docs/data_preparation.md).
## Train
You can use the following commands to train a model.
For more details and optional arguments infos, you can refer to **Training setting** part in [getting_started](/docs/getting_started.md#training-setting).
## Inference
You can use the following commands to inference a model.
1. (Optional) You can use the following command to generate a formatted proposal file, which will be fed into the action classifier (Currently supports only SSN and P-GCN, not including TSN, I3D etc.) to get the classification result of proposals.
[Temporal Action Detection With Structured Segment Networks](https://openaccess.thecvf.com/content_iccv_2017/html/Zhao_Temporal_Action_Detection_ICCV_2017_paper.html)
<!-- [ALGORITHM] -->
## Abstract
<!-- [ABSTRACT] -->
Detecting actions in untrimmed videos is an important yet challenging task. In this paper, we present the structured segment network (SSN), a novel framework which models the temporal structure of each action instance via a structured temporal pyramid. On top of the pyramid, we further introduce a decomposed discriminative model comprising two classifiers, respectively for classifying actions and determining completeness. This allows the framework to effectively distinguish positive proposals from background or incomplete ones, thus leading to both accurate recognition and localization. These components are integrated into a unified network that can be efficiently trained in an end-to-end fashion. Additionally, a simple yet effective temporal action proposal scheme, dubbed temporal actionness grouping (TAG) is devised to generate high quality action proposals. On two challenging benchmarks, THUMOS14 and ActivityNet, our method remarkably outperforms previous state-of-the-art methods, demonstrating superior accuracy and strong adaptivity in handling actions with various temporal structures.
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. Since SSN utilizes different structured temporal pyramid pooling methods at training and testing, please refer to [ssn_r50_450e_thumos14_rgb_train](/configs/localization/ssn/ssn_r50_450e_thumos14_rgb_train.py) at training and [ssn_r50_450e_thumos14_rgb_test](/configs/localization/ssn/ssn_r50_450e_thumos14_rgb_test.py) at testing.
3. We evaluate the action detection performance of SSN, using action proposals of TAG. For more details on data preparation, you can refer to thumos14 TAG proposals in [Data Preparation](/docs/data_preparation.md).
4. The reference SSN in is evaluated with `ResNet50` backbone in MMAction, which is the same backbone with ours. Note that the original setting of MMAction SSN uses the `BNInception` backbone.
:::
## Train
You can use the following command to train a model.
For more details and optional arguments infos, you can refer to **Training setting** part in [getting_started](/docs/getting_started.md#training-setting).
## Test
You can use the following command to test a model.
