Dynamic Facial Expression Recognition under Partial Occlusion with Optical Flow Reconstruction

TL;DR

This paper introduces a novel optical flow reconstruction method using auto-encoders with skip connections to improve facial expression recognition under partial occlusion, demonstrating significant accuracy gains on the CK+ dataset.

Contribution

It proposes the first direct movement reconstruction approach for facial expression recognition under occlusion using optical flow and auto-encoders.

Findings

Significant reduction in recognition accuracy gap between occluded and non-occluded faces.

Outperforms existing state-of-the-art occlusion handling methods.

Introduces a new experimental protocol for occlusion generation and reconstruction evaluation.

Abstract

Video facial expression recognition is useful for many applications and received much interest lately. Although some solutions give really good results in a controlled environment (no occlusion), recognition in the presence of partial facial occlusion remains a challenging task. To handle occlusions, solutions based on the reconstruction of the occluded part of the face have been proposed. These solutions are mainly based on the texture or the geometry of the face. However, the similarity of the face movement between different persons doing the same expression seems to be a real asset for the reconstruction. In this paper we exploit this asset and propose a new solution based on an auto-encoder with skip connections to reconstruct the occluded part of the face in the optical flow domain. To the best of our knowledge, this is the first proposition to directly reconstruct the movement for facial expression recognition. We validated our approach in the controlled dataset CK+ on which different occlusions were generated. Our experiments show that the proposed method reduce significantly the gap, in terms of recognition accuracy, between occluded and non-occluded situations. We also compare our approach with existing state-of-the-art solutions. In order to lay the basis of a reproducible and fair comparison in the future, we also propose a new experimental protocol that includes occlusion generation and reconstruction evaluation.