Source code for the following paper(arXiv link):

    Learning Actor Relation Graphs for Group Activity Recognition
    Jianchao Wu, Limin Wang, Li Wang, Jie Guo, Gangshan Wu
    in CVPR 2019

• Python 3.x
• PyTorch 0.4.1
• numpy, pickle, scikit-image
• RoIAlign for Pytorch
• Datasets: Volleyball, Collective

1. Download volleyball or collective dataset file.
2. Unzip the dataset file into data/volleyball or data/collective.

1. Stage1: Fine-tune the model on single frame without using GCN.

   # volleyball dataset
   python scripts/train_volleyball_stage1.py
   
   # collective dataset
   python scripts/train_collective_stage1.py

2. Stage2: Fix weights of the feature extraction part of network, and train the network with GCN.

   # volleyball dataset
   python scripts/train_volleyball_stage2.py
   
   # collective dataset
   python scripts/train_collective_stage2.py

   You can specify the running arguments in the python files under scripts/ directory. The meanings of arguments can be found in config.py

@inproceedings{CVPR2019_ARG,
  title = {Learning Actor Relation Graphs for Group Activity Recognition},
  author = {Jianchao Wu and Limin Wang and Li Wang and Jie Guo and Gangshan Wu},
  booktitle = {CVPR},
  year = {2019},
}