Facial Expression Recognition Using Residual Masking Network

TL;DR

This paper introduces a Residual Masking Network that combines residual and U-Net architectures with a segmentation-based attention mechanism to improve facial expression recognition accuracy, achieving state-of-the-art results on benchmark datasets.

Contribution

It presents a novel Residual Masking Network that enhances CNN performance in FER by integrating segmentation-based feature refinement, a new approach in deep facial expression recognition.

Findings

Achieved state-of-the-art accuracy on FER2013 dataset.

Outperformed existing methods on private VEMO dataset.

Demonstrated effectiveness of segmentation-based attention in FER.

Abstract

Automatic facial expression recognition (FER) has gained much attention due to its applications in human-computer interaction. Among the approaches to improve FER tasks, this paper focuses on deep architecture with the attention mechanism. We propose a novel Masking idea to boost the performance of CNN in facial expression task. It uses a segmentation network to refine feature maps, enabling the network to focus on relevant information to make correct decisions. In experiments, we combine the ubiquitous Deep Residual Network and Unet-like architecture to produce a Residual Masking Network. The proposed method holds state-of-the-art (SOTA) accuracy on the well-known FER2013 and private VEMO datasets. The source code is available at https://github.com/phamquiluan/ResidualMaskingNetwork.