---
license: mit
---

# ekman_expressions

This repository contains all the models trained in the scientific article titled *"Unveiling the Human-like Similarities of Automatic Facial Expression Recognition: An Empirical Exploration through Explainable AI"*.

To reproduce the results of the article, please refer to the Github project containing all code used: [https://github.com/Xavi3398/ekman_expressions](https://github.com/Xavi3398/ekman_expressions).


## Datasets used for training:
The datasets used were 5: CK+, BU-4DFE, JAFFE and WSEFEP, and FEGA. These are some examples from each class:
<img src="resources/datasets.png" alt="datasets" style="width:60%;height:auto;display:block;margin-left:auto;margin-right:auto;">


## Models:
We explored 12 different Deep Learning models:

| Model          | Image Size | Pre-training | Parameters |
|----------------------|---------------------|-----------------------|---------------------|
| AlexNet        | 224x224             | No                    | 88.7 M              |
| WeiNet         | 64x64               | No                    | 1.7 M               |
| SongNet        | 224x224             | No                    | 172.7 K             |
| SilNet         | 150x150             | No                    | 184.9 M             |
| VGG16          | 224x224             | Yes                   | 14.7 M              |
| VGG19          | 224x224             | Yes                   | 20 M                |
| ResNet50       | 224x224             | Yes                   | 23.6 M              |
| ResNet101V2    | 224x224             | Yes                   | 42.6 M              |
| InceptionV3    | 224x224             | Yes                   | 21.8 M              |
| Xception       | 224x224             | Yes                   | 20.9 M              |
| MobileNetV3    | 224x224             | Yes                   | 3 M                 |
| EfficientNetV2 | 224x224             | Yes                   | 5.9 M               |


## License

This project is licensed under the terms of the MIT license. See the [LICENSE](LICENSE) file for details.


## Acknowledgments
Grant PID2019-104829RA-I00 funded by MCIN/ AEI /10.13039/501100011033. Project EXPLainable Artificial INtelligence systems for health and well-beING (EXPLAINING)

This work is part of the Project PID2022-136779OB-C32 (PLEISAR) funded by MICIU/ AEI /10.13039/501100011033/ and FEDER, EU. Project Playful Experiences with Interactive Social Agents and Robots (PLEISAR): Social Learning and Intergenerational Communication.

F. X. Gaya-Morey was supported by an FPU scholarship from the Ministry of European Funds, University and Culture of the Government of the Balearic Islands.


## Citation

If you use this code in your research, please cite our paper:

```
@misc{gayamorey2024unveilinghumanlikesimilaritiesautomatic,
      title={Unveiling the Human-like Similarities of Automatic Facial Expression Recognition: An Empirical Exploration through Explainable AI}, 
      author={F. Xavier Gaya-Morey and Silvia Ramis-Guarinos and Cristina Manresa-Yee and Jose M. Buades-Rubio},
      year={2024},
      eprint={2401.11835},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2401.11835}, 
}
```


## Contact

If you have any questions or feedback, please feel free to contact the authors.