This is an official implementation of DECOMPL: Decompositional Learning with Attention Pooling for Group Activity Recognition from a Single Volleyball Image. In this repository, we provide PyTorch code for training and testing as described in the paper.
data
and set its name to volleyball_videos
tracks_normalized.pkl
from cvlab-epfl/social-scene-understanding and put it into the directory data
Conda (Recommended):
conda create -n DECOMPL
conda activate DECOMPL
Pip
pip install -r requirements.txt
Training and Validation: Modify test_only
argument in /scripts/run_model_volleyball.py
to train or validate. To use the pretrained weights set load_path
to checkpoint_weights_volleyball_half.pth
cd PROJECT_PATH
python scripts/run_model.py
Additionally, Training and Validation for Collective Activity Dataset: Follow the same instructions as for the Volleyball Dataset on the scripts /scripts/run_model_collective.py
and pretrained weights checkpoint_weights_collective_half.pth
Note: The weights provided are converted to half precision due to size constraints.
@article{demirel2023decompl,
title={DECOMPL: Decompositional Learning with Attention Pooling for Group Activity Recognition from a Single Volleyball Image},
author={Demirel, Berker and Ozkan, Huseyin},
journal={arXiv preprint arXiv:2303.06439},
year={2023}
}