OpenTAD: An Open-Source Temporal Action Detection Toolbox.

OpenTAD is an open-source temporal action detection (TAD) toolbox based on PyTorch.

🥳 What's New

• A technical report of this library will be provided soon.
• [2024/06/06] 🔥 We release version v0.1.2, which brings many new features and improvements.
• [2024/06/01] 🔥 We rank 1st in the Action Recognition, Action Detection, and Audio-Based Interaction Detection tasks of the EPIC-KITCHENS-100 2024 Challenge, as well as 1st place in the Moment Queries task of the Ego4D 2024 Challenge by using OpenTAD! Code will be released soon!
• [2024/04/17] We release the AdaTAD, which can achieve average mAP of 42.90% on ActivityNet and 77.07% on THUMOS14.

📖 Major Features

• Support SoTA TAD methods with modular design. We decompose the TAD pipeline into different components, and implement them in a modular way. This design makes it easy to implement new methods and reproduce existing methods.
• Support multiple TAD datasets. We support 8 TAD datasets, including ActivityNet-1.3, THUMOS-14, HACS, Ego4D-MQ, Epic-Kitchens-100, FineAction, Multi-THUMOS, Charades datasets.
• Support feature-based training and end-to-end training. The feature-based training can easily be extended to end-to-end training with raw video input, and the video backbone can be easily replaced.
• Release various pre-extracted features. We release the feature extraction code, as well as many pre-extracted features on each dataset.

🌟 Model Zoo

One Stage	Two Stage	DETR	End-to-End Training
ActionFormer (ECCV'22) TriDet (CVPR'23) TemporalMaxer (arXiv'23) VideoMambaSuite (arXiv'24)	BMN (ICCV'19) GTAD (CVPR'20) TSI (ACCV'20) VSGN (ICCV'21)	TadTR (TIP'22)	AFSD (CVPR'21) E2E-TAD (CVPR'22) ETAD (CVPRW'23) Re2TAL (CVPR'23) AdaTAD (CVPR'24)

The detailed configs, results, and pretrained models of each method can be found in above folders.

🛠️ Installation

Please refer to install.md for installation and data preparation.

🚀 Usage

Please refer to usage.md for details of training and evaluation scripts.

📄 Updates

Please refer to changelog.md for update details.

🤝 Roadmap

All the things that need to be done in the future is in roadmap.md.

🖊️ Citation

[Acknowledgement] This repo is inspired by OpenMMLab project, and we give our thanks to their contributors.

If you think this repo is helpful, please cite us:

@misc{2024opentad,
    title={OpenTAD: An Open-Source Toolbox for Temporal Action Detection},
    author={Shuming Liu, Chen Zhao, Fatimah Zohra, Mattia Soldan, Carlos Hinojosa, Alejandro Pardo, Anthony Cioppa, Lama Alssum, Mengmeng Xu, Merey Ramazanova, Juan León Alcázar, Silvio Giancola, Bernard Ghanem},
    howpublished = {\url{https://github.com/sming256/opentad}},
    year={2024}
}

If you have any questions, please contact: shuming.liu@kaust.edu.sa.