Solution for EPiC 2023 competition by VSL team - Pattern Recognition Lab, Chonnam Nat'l Univ.
- Create conda environment
conda create -n epic_vsl python=3.9
conda activate epic_vsl
pip install -r requirements.txt
- To train the model, edit OUT_DIR and DATA_DIR in
config/base_cfg.yaml, or directly inscripts/*_train.sh. The scripts for training our 3 attempts are in scripts. The pretrained weights, prediction, completed configs of our 3 attempts could be found in this link.
Our repo are trained on Debian 11.5 with CUDA 12.1 and CUDNN 8.
@article{vu2023multi,
title={Multi-scale Transformer-based Network for Emotion Recognition from Multi Physiological Signals},
author={Vu, Tu and Huynh, Van Thong and Kim, Soo-Hyung},
journal={arXiv preprint arXiv:2305.00769},
year={2023}
}