Learning to Amend Facial Expression Representation via De-albino and Affinity

TL;DR

This paper introduces the Amending Representation Module (ARM), a novel architecture that improves facial expression recognition by addressing padding erosion and feature weakening, achieving state-of-the-art results on multiple benchmarks.

Contribution

The paper proposes ARM, a new module that replaces pooling layers to enhance facial expression features and mitigate padding erosion effects in FER models.

Findings

ARM improves FER accuracy on RAF-DB, Affect-Net, and SFEW datasets.

ARM outperforms existing state-of-the-art methods.

The approach effectively reduces feature erosion and simplifies representation learning.

Abstract

Facial Expression Recognition (FER) is a classification task that points to face variants. Hence, there are certain affinity features between facial expressions, receiving little attention in the FER literature. Convolution padding, despite helping capture the edge information, causes erosion of the feature map simultaneously. After multi-layer filling convolution, the output feature map named albino feature definitely weakens the representation of the expression. To tackle these challenges, we propose a novel architecture named Amending Representation Module (ARM). ARM is a substitute for the pooling layer. Theoretically, it can be embedded in the back end of any network to deal with the Padding Erosion. ARM efficiently enhances facial expression representation from two different directions: 1) reducing the weight of eroded features to offset the side effect of padding, and 2) decomposing facial features to simplify representation learning. Experiments on public benchmarks prove that our ARM boosts the performance of FER remarkably. The validation accuracies are respectively 90.42% on RAF-DB, 65.2% on Affect-Net, and 58.71% on SFEW, exceeding current state-of-the-art methods. Our implementation and trained models are available at https://github.com/JiaweiShiCV/Amend-Representation-Module.