PyTorch implementation of the article "You Only Watch Once: A Unified CNN Architecture for Real-Time Spatiotemporal Action Localization". The repositry contains code for real-time spatiotemporal action localization with PyTorch on AVA, UCF101-24 and JHMDB datasets!
Updated paper can be accessed via YOWO_updated.pdf
AVA dataset visualizations!
UCF101-24 and J-HMDB-21 datasets visualizations!
In this work, we present YOWO (You Only Watch Once), a unified CNN architecture for real-time spatiotemporal action localization in video stream. YOWO is a single-stage framework, the input is a clip consisting of several successive frames in a video, while the output predicts bounding box positions as well as corresponding class labels in current frame. Afterwards, with specific strategy, these detections can be linked together to generate Action Tubes in the whole video.
Since we do not separate human detection and action classification procedures, the whole network can be optimized by a joint loss in an end-to-end framework. We have carried out a series of comparative evaluations on two challenging representative datasets UCF101-24 and J-HMDB-21. Our approach outperforms the other state-of-the-art results while retaining real-time capability, providing 34 frames-per-second on 16-frames input clips and 62 frames-per-second on 8-frames input clips.
git clone https://github.com/wei-tim/YOWO.git
cd YOWOUse instructions here for the preperation of AVA dataset.
Modify the paths in ucf24.data and jhmdb21.data under cfg directory accordingly. Download the dataset annotations from here.
- Darknet-19 weights can be downloaded via:
wget http://pjreddie.com/media/files/yolo.weights- ResNeXt ve ResNet pretrained models can be downloaded from here.
NOTE: For JHMDB-21 trainings, HMDB-51 finetuned pretrained models should be used! (e.g. "resnext-101-kinetics-hmdb51_split1.pth").
- For resource efficient 3D CNN architectures (ShuffleNet, ShuffleNetv2, MobileNet, MobileNetv2), pretrained models can be downloaded from here.
Pretrained models for UCF101-24 and J-HMDB-21 datasets can be downloaded from here.
Pretrained models for AVA dataset can be downloaded from here.
All materials (annotations and pretrained models) are also available in Baiduyun Disk: here with password 95mm
- All training configurations are given in ava.yaml, ucf24.yaml and jhmdb.yaml files.
- AVA training:
python main.py --cfg cfg/ava.yaml- UCF101-24 training:
python main.py --cfg cfg/ucf24.yaml- J-HMDB-21 training:
python main.py --cfg cfg/jhmdb.yaml-
For AVA dataset, after each epoch, validation is performed and frame-mAP score is provided.
-
For UCF101-24 and J-HMDB-21 datasets, after each validation, frame detections is recorded under 'jhmdb_detections' or 'ucf_detections'. From here, 'groundtruths_jhmdb.zip' and 'groundtruths_jhmdb.zip' should be downloaded and extracted to "evaluation/Object-Detection-Metrics". Then, run the following command to calculate frame_mAP.
python evaluation/Object-Detection-Metrics/pascalvoc.py --gtfolder PATH-TO-GROUNDTRUTHS-FOLDER --detfolder PATH-TO-DETECTIONS-FOLDER
- For video_mAP, set the pretrained model in the correct yaml file and run:
python video_mAP.py --cfg cfg/ucf24.yaml- You can run AVA pretrained model on any test video with the following code:
python test_video_ava.py --cfg cfg/ava.yamlUPDATEs:
- YOWO is extended for AVA dataset.
- Old repo is deprecated and moved to YOWO_deprecated branch.
If you use this code or pre-trained models, please cite the following:
@InProceedings{kopuklu2019yowo,
title={You Only Watch Once: A Unified CNN Architecture for Real-Time Spatiotemporal Action Localization},
author={K{\"o}p{\"u}kl{\"u}, Okan and Wei, Xiangyu and Rigoll, Gerhard},
journal={arXiv preprint arXiv:1911.06644},
year={2019}
}We thank Hang Xiao for releasing pytorch_yolo2 codebase, which we build our work on top.