R2-Trans:Fine-Grained Visual Categorization with Redundancy Reduction

TL;DR

R2-Trans introduces a novel FGVC method that reduces redundancy in class tokens and adaptively extracts discriminative regions, leading to improved accuracy on benchmark datasets.

Contribution

The paper proposes a new approach combining adaptive masking and the Information Bottleneck to enhance fine-grained visual categorization performance.

Findings

Outperforms state-of-the-art methods on benchmark datasets

Effectively reduces redundant information in class tokens

Improves discriminative region extraction accuracy

Abstract

Fine-grained visual categorization (FGVC) aims to discriminate similar subcategories, whose main challenge is the large intraclass diversities and subtle inter-class differences. Existing FGVC methods usually select discriminant regions found by a trained model, which is prone to neglect other potential discriminant information. On the other hand, the massive interactions between the sequence of image patches in ViT make the resulting class-token contain lots of redundant information, which may also impacts FGVC performance. In this paper, we present a novel approach for FGVC, which can simultaneously make use of partial yet sufficient discriminative information in environmental cues and also compress the redundant information in class-token with respect to the target. Specifically, our model calculates the ratio of high-weight regions in a batch, adaptively adjusts the masking threshold and achieves moderate extraction of background information in the input space. Moreover, we also use the Information Bottleneck~(IB) approach to guide our network to learn a minimum sufficient representations in the feature space. Experimental results on three widely-used benchmark datasets verify that our approach can achieve outperforming performance than other state-of-the-art approaches and baseline models.