RefineLoc: Iterative Refinement for Weakly-Supervised Action Localization

[Paper] [Project Website]

This repository holds the official implementation of RefineLoc method presented in WACV 2021.

RefineLoc: Iterative Refinement for Weakly-Supervised Action Localization. Alejandro Pardo*, Humam Alwassel*, Fabian Caba Heilbron, Ali Thabet, Bernard Ghanem. In WACV, 2021.

Installation

Create the conda environment.

conda env create -f environment.yml

Download the features from the links provided in data/README.md and place them in the correct subfolders inside the data folder.

Training

Run the following command to reproduce the ActivityNet results presented in the paper:

sh src/slurm_scripts/slurm_run_best.sh

To reproduce the THUMOS14 results, change CONFIG_TYPE to best_thumos14 in src/slurm_scripts/slurm_run_best.sh.

RefineLoc + WTAL and RefineLoc + BasNet

The two repos will be available soon.

Please cite this work if you find the code useful for your research.

@InProceedings{pardo_2021_refineloc,
  title={RefineLoc: Iterative Refinement for Weakly-Supervised Action Localization},
  author={Pardo, Alejandro and Alwassel, Humam and Heilbron, Fabian Caba and 
          Thabet, Ali and Ghanem, Bernard},
  booktitle={Proceedings of the IEEE/CVF Winter Conference on Applications of 
             Computer Vision (WACV)},
  year={2021}
}