Vision Transformer Equipped with Neural Resizer on Facial Expression Recognition Task

TL;DR

This paper introduces Neural Resizer, a data-driven approach to improve Vision Transformers for facial expression recognition in low-quality, imbalanced data scenarios, achieving near state-of-the-art results.

Contribution

The paper proposes Neural Resizer, a novel training framework that enhances Vision Transformers by compensating for low-resolution data in facial expression recognition tasks.

Findings

Neural Resizer improves Transformer performance on facial expression data.

The approach nearly achieves state-of-the-art accuracy.

Data-driven resizing outperforms traditional interpolation methods.

Abstract

When it comes to wild conditions, Facial Expression Recognition is often challenged with low-quality data and imbalanced, ambiguous labels. This field has much benefited from CNN based approaches; however, CNN models have structural limitation to see the facial regions in distant. As a remedy, Transformer has been introduced to vision fields with global receptive field, but requires adjusting input spatial size to the pretrained models to enjoy their strong inductive bias at hands. We herein raise a question whether using the deterministic interpolation method is enough to feed low-resolution data to Transformer. In this work, we propose a novel training framework, Neural Resizer, to support Transformer by compensating information and downscaling in a data-driven manner trained with loss function balancing the noisiness and imbalance. Experiments show our Neural Resizer with F-PDLS loss function improves the performance with Transformer variants in general and nearly achieves the state-of-the-art performance.