Bidirectional Logits Tree: Pursuing Granularity Reconcilement in Fine-Grained Classification

TL;DR

This paper introduces the Bidirectional Logits Tree framework for fine-grained classification, enabling better semantic reconciliation across granularities by sequentially learning classifiers from fine to coarse levels, and incorporates intra-granularity difference learning.

Contribution

The paper proposes a novel bidirectional logits tree framework with sequential classifier learning and intra-granularity difference learning for improved fine-grained classification.

Findings

Effective in reconciling semantic gaps across granularities

Outperforms existing methods on benchmark datasets

Enhances subtle class distinction within the same granularity

Abstract

This paper addresses the challenge of Granularity Competition in fine-grained classification tasks, which arises due to the semantic gap between multi-granularity labels. Existing approaches typically develop independent hierarchy-aware models based on shared features extracted from a common base encoder. However, because coarse-grained levels are inherently easier to learn than finer ones, the base encoder tends to prioritize coarse feature abstractions, which impedes the learning of fine-grained features. To overcome this challenge, we propose a novel framework called the Bidirectional Logits Tree (BiLT) for Granularity Reconcilement. The key idea is to develop classifiers sequentially from the finest to the coarsest granularities, rather than parallelly constructing a set of classifiers based on the same input features. In this setup, the outputs of finer-grained classifiers serve as inputs for coarser-grained ones, facilitating the flow of hierarchical semantic information across different granularities. On top of this, we further introduce an Adaptive Intra-Granularity Difference Learning (AIGDL) approach to uncover subtle semantic differences between classes within the same granularity. Extensive experiments demonstrate the effectiveness of our proposed method.