Learning Contrastive Self-Distillation for Ultra-Fine-Grained Visual Categorization Targeting Limited Samples

TL;DR

This paper introduces CSDNet, a novel framework combining contrastive learning and self-distillation to improve ultra-fine-grained visual categorization with limited data, achieving superior performance.

Contribution

The work presents a new method with modules for discrepancy parsing, dynamic learning, and self-distillation tailored for Ultra-FGVC, enhancing model generalization and discrimination.

Findings

CSDNet outperforms existing Ultra-FGVC methods.

Adaptive augmentation improves subcategory discrepancy focus.

Dynamic memory enhances feature learning and contrastive training.

Abstract

In the field of intelligent multimedia analysis, ultra-fine-grained visual categorization (Ultra-FGVC) plays a vital role in distinguishing intricate subcategories within broader categories. However, this task is inherently challenging due to the complex granularity of category subdivisions and the limited availability of data for each category. To address these challenges, this work proposes CSDNet, a pioneering framework that effectively explores contrastive learning and self-distillation to learn discriminative representations specifically designed for Ultra-FGVC tasks. CSDNet comprises three main modules: Subcategory-Specific Discrepancy Parsing (SSDP), Dynamic Discrepancy Learning (DDL), and Subcategory-Specific Discrepancy Transfer (SSDT), which collectively enhance the generalization of deep models across instance, feature, and logit prediction levels. To increase the diversity of training samples, the SSDP module introduces adaptive augmented samples to spotlight subcategory-specific discrepancies. Simultaneously, the proposed DDL module stores historical intermediate features by a dynamic memory queue, which optimizes the feature learning space through iterative contrastive learning. Furthermore, the SSDT module effectively distills subcategory-specific discrepancies knowledge from the inherent structure of limited training data using a self-distillation paradigm at the logit prediction level. Experimental results demonstrate that CSDNet outperforms current state-of-the-art Ultra-FGVC methods, emphasizing its powerful efficacy and adaptability in addressing Ultra-FGVC tasks.