Awesome
A Sliding Window Scheme for Online Temporal Action Localization (OAT-OSN)
A Sliding Window Scheme for Online Temporal Action Localization (ECCV2022)
Young Hwi Kim, Hyolim Kang, Seon Joo Kim
[link]
Updates
17 Jul, 2022: Initial release
Installation
Prerequisites
- Ubuntu 18.04
- Python 3.8.8
- CUDA 11.0
Requirements
- pytorch==1.8.1
- numpy==1.19.2
- h5py==3.6.0
- ...
To install all required libraries, execute the pip command below.
pip install -r requirement.txt
Training
Input Features
We provide the Kinetics pre-trained feature of THUMOS'14 dataset.
The extracted features can be downloaded from link.
Files should be located in 'data/'.
You can also get the feature files from here.
Trained Model
The trained models that used Kinetics pre-trained feature can be downloaded from link.
Files should be located in 'checkpoints/'.
Training Model by own
To train the main OAT model, execute the command below.
python main.py --mode=train
To train the post-processing network (OSN), execute the commands below.
python supnet.py --mode=make --inference_subset=train
python supnet.py --mode=make --inference_subset=test
python supnet.py --mode=train
Testing
To test OAT-OSN, execute the command below.
python main.py --mode=test
To test OAT-NMS, execute the command below.
python main.py --mode=test --pptype=nms
Results
| Method | 0.3 | 0.4 | 0.5 | 0.6 | 0.7 |
|---|---|---|---|---|---|
| OAT-OSN | 63.0 | 56.7 | 47.1 | 36.3 | 20.0 |
| OAT-NMS | 69.7 | 64.0 | 53.9 | 42.9 | 27.0 |
Citing OAT-OSN
Please cite our paper in your publications if it helps your research:
@inproceedings{kim2022sliding,
title={A Sliding Window Scheme for Online Temporal Action Localization},
author={Kim, Young Hwi and Kang, Hyolim and Kim, Seon Joo},
booktitle={European Conference on Computer Vision},
pages={653--669},
year={2022},
organization={Springer}
}