To Start Automatic Commentary of Soccer Game with Mixed Spatial and Temporal Attention

How to create environment required

Run the following command on terminal

conda create -n MixSaT  -c conda-forge pytorch-gpu=1.13 pytorch-lightning=1.7 torchmetrics==0.11.4 python=3.10 einops scikit-learn
conda activate MixSaT
pip install wandb SoccerNet rich

The wandb account must be configured in advance to proper output log to wandb experiment logger. More details can be refered to Wandb

How to train

python main.py 

How to interference with pretrained checkpoint

Download the pretrained model on OneDrive, and run following command on terminal

python main.py --test_only --ckpt_path [LOCATION OF CHECKPOINT]

Citation

If you find our work useful for your research, please do not hesitate to cite our paper

@INPROCEEDINGS{9978078,
  author={Chan, Cheuk-Yiu and Hui, Chun-Chuen and Siu, Wan-Chi and Chan, Sin-wai and Chan, H. Anthony},
  booktitle={TENCON 2022 - 2022 IEEE Region 10 Conference (TENCON)}, 
  title={To Start Automatic Commentary of Soccer Game with Mixed Spatial and Temporal Attention}, 
  year={2022},
  volume={},
  number={},
  pages={1-6},
  doi={10.1109/TENCON55691.2022.9978078}}