This is code for our submission in the expression track of ABAW 2022 competition as a part of ECCV 2022.
Title: Semi-supervised Multi-task Facial Affect Recongition
Abstract:Automatic affect recognition has applications in many areas such as education, gaming, software development, automotives, medical care, etc. but it is non trivial task to achieve appreciable perfor- mance on in-the-wild data sets. In-the-wild data sets though represent real-world scenarios better than synthetic data sets, the former ones suffer from the problem of incomplete labels. Inspired by semi-supervised learning, in this paper, we introduce our submission to the Multi-Task-Learning Challenge at the 4th Affective Behavior Analysis in-the-wild (ABAW) 2022 Competition. The three tasks that are considered in this challenge are valence-arousal(VA) estimation, classification of expressions into 6 basic (anger, disgust, fear, happiness, sadness, surprise), neutral, and the ’other’ category and 12 action units(AU) numbered AU-{1,2,4,6,7,10,12,15,23,24,25,26}. Our method Semi-supervised Multi-task Facial Affect Recognition titled SS-MFAR uses a deep residual net- work with task specific classifiers for each of the tasks along with adaptive thresholds for each expression class and semi-supervised learning for the incomplete labels.
Link to the paper : http://arxiv.org/abs/2207.09012
Link to checkpoints
https://drive.google.com/drive/folders/1YWNP6dFBvLQhZHVcr1wUgKRePv5rIMCJ?usp=sharing