Balancing methods for single-label data cannot be applied to multi-label problems as they would also resample the samples with high occurrences. We propose to reformulate this problem as an optimization problem in order to balance multi-label data. We apply this balancing algorithm to training datasets for detecting isolated facial movements, so-called Action Units. Several Action Units can describe combined emotions or physical states such as pain. As datasets in this area are limited and mostly imbalanced, we show how optimized balancing and then augmentation can improve Action Unit detection. At the IEEE Conference on Face and Gesture Recognition 2020, we ranked third in the Affective Behavior Analysis in-the-wild (ABAW) challenge for the Action Unit detection task.
@article{arxiv.2003.08751,
title = {Unique Class Group Based Multi-Label Balancing Optimizer for Action Unit Detection},
author = {Ines Rieger and Jaspar Pahl and Dominik Seuss},
journal= {arXiv preprint arXiv:2003.08751},
year = {2021}
}
Comments
Accepted at the 15th IEEE International Conference on Automatic Face and Gesture Recognition 2020, Workshop "Affect Recognition in-the-wild: Uni/Multi-Modal Analysis & VA-AU-Expression Challenges". arXiv admin note: substantial text overlap with arXiv:2002.03238