FER SSL

Official Of our ACII 2023 paper:

Active Learning with Contrastive Pre-training for Facial Expression Recognition
Shuvendu Roy, Ali Etemad
In Proceedings of the IEEE International Conference on Affective Computing and Intelligent Interaction (ACII), 2023

Dataset

We used the following dataset in this work. Plese follow the instructions in the respective website to download the dataset.

Store the datasets in the ./data directory. The processed dataset structure should look like this:

dataset_name
├── train
│   ├── class_001
|   |      ├── 1.jpg
|   |      ├── 2.jpg
|   |      └── ...
│   ├── class_002
|   |      ├── 1.jpg
|   |      ├── 2.jpg
|   |      └── ...
│   └── ...
└── val
    ├── class_001
    |      ├── 1.jpg
    |      ├── 2.jpg
    |      └── ...
    ├── class_002
    |      ├── 1.jpg
    |      ├── 2.jpg
    |      └── ...
    └── ...

Installation

Make sure you install the requirements before trying to run the models.

pip3 install -r requirements/requirements.txt

Training

The training process of the proposed solution involves two steps. First pre-train the encoder with the following script.

python pre_train.py --dataset FER13 \
  --batch_size 1024 \
  --learning_rate 0.5 \
  --temp 0.1 \
  --cosine

Then, train the model with the following script.

python train.py --config_path config/FER13.yaml

make sure to replace the path to the pre-trained model in the config file.

Results

Acknowledgements

We would like to thank the authors of the following repositories for releasing their code. We used their code as a starting point for our implementation for active learning methods.

https://github.com/decile-team/distil

Citation

Please cite our paper if you this code repo in your work.

@inproceedings{roy2023active,
  title={Active Learning with Contrastive Pre-training for Facial Expression Recognition},
  author={Roy, Shuvendu and Etemad, Ali},
  booktitle={11th International Conference on Affective Computing and Intelligent Interaction (ACII)},
  pages={1--8},
  year={2023},
  organization={IEEE}
}

Contact

Thanks for your attention! If you have any suggestion or question, you can contact us:

shuvendu.roy@queensu.ca

drpengsong/ActiveFER