Our work has been published in Pattern Recognition letters.

https://authors.elsevier.com/a/1ca-7cAmylz0f

https://www.sciencedirect.com/science/article/abs/pii/S0167865521000489#absh0002

Title: Landmark Guidance Independent Spatio-channel Attention and Complementary Context Information based Facial Expression Recognition

https://arxiv.org/pdf/2007.10298v1.pdf

This work was presented in FG-2020 workshop: Affect Recognition in-the-wild: Uni/Multi-Modal Analysis & VA-AU-Expression Challenges

IEEE Face and Gesture Recognition (FG-2020) Competition: Affective Behavior Analysis in-the-wild (ABAW)

Leaderboard: https://ibug.doc.ic.ac.uk/resources/fg-2020-competition-affect

Darshan Gera and S. Balasubramanian."Affect Expression Behaviour Analysis in the Wild (ABAW) using Spatio-channel Attention and Complementary Context Information", Arxiv preprint: http://arxiv.org/abs/2009.14440

Source Code: https://github.com/1980x/ABAW2020DMACS

Our proposed FER framework:

Activation maps using GRADCAM for few of FEDRO images are shown below:

Example of our synthetic created masked face images from FERPLUS dataset are shown below:

Comparision of activation maps using GRADCAM between Baseline3, RAN and Our model for masked face images on FERPlus are shown below:

Note: Our trained model checkpoints will be made available after publication.

Citation: @article{gera2020landmark, title={Landmark Guidance Independent Spatio-channel Attention and Complementary Context Information based Facial Expression Recognition}, author={Gera, Darshan and Balasubramanian, S}, journal={arXiv preprint arXiv:2007.10298}, year={2020} }

Acknowledgments: We dedicate this work to Bhagawan Sri Sathya Sai Baba, Divine Founder Chancellor of Sri Sathya Sai Institute of Higher Learning, PrasanthiNilyam, A.P., India.