Learning Part Knowledge to Facilitate Category Understanding for Fine-Grained Generalized Category Discovery

2025-03-21

AI Summary

To address the challenge in fine-grained generalized category discovery (GCD) where global features fail to capture local discriminative differences, this paper proposes an unsupervised adaptive part learning framework. The method introduces three key innovations: (1) GMM-driven adaptive part decomposition, enabling automatic discovery of semantically meaningful local regions without part-level annotations; (2) part-level contrastive regularization, explicitly enhancing the discriminability of local features; and (3) a global–local collaborative representation learning mechanism that jointly optimizes holistic and part-based embeddings. Evaluated on multiple fine-grained benchmarks, the approach achieves state-of-the-art performance. Moreover, it demonstrates strong generalization capability on standard (non-fine-grained) GCD datasets, confirming both effectiveness and robustness. The framework bridges the gap between coarse-grained and fine-grained GCD by unifying unsupervised part discovery with contrastive representation learning.

Technology Category

• Machine Learning: Unsupervised & Self-Supervised Learning
• Computer Vision: Representation Learning for Vision
• Search and Optimization: Local Search

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Generalized Category Discovery (GCD) aims to classify unlabeled data containing both seen and novel categories. Although existing methods perform well on generic datasets, they struggle in fine-grained scenarios. We attribute this difficulty to their reliance on contrastive learning over global image features to automatically capture discriminative cues, which fails to capture the subtle local differences essential for distinguishing fine-grained categories. Therefore, in this paper, we propose incorporating part knowledge to address fine-grained GCD, which introduces two key challenges: the absence of annotations for novel classes complicates the extraction of the part features, and global contrastive learning prioritizes holistic feature invariance, inadvertently suppressing discriminative local part patterns. To address these challenges, we propose PartGCD, including 1) Adaptive Part Decomposition, which automatically extracts class-specific semantic parts via Gaussian Mixture Models, and 2) Part Discrepancy Regularization, enforcing explicit separation between part features to amplify fine-grained local part distinctions. Experiments demonstrate state-of-the-art performance across multiple fine-grained benchmarks while maintaining competitiveness on generic datasets, validating the effectiveness and robustness of our approach.

Problem

• Classify unlabeled data with seen and novel categories in fine-grained scenarios
• Extract part features without annotations for novel classes
• Enhance discriminative local part patterns in contrastive learning

Innovation

• Adaptive Part Decomposition via Gaussian Mixture Models
• Part Discrepancy Regularization for local distinctions
• Learning Part Knowledge for fine-grained GCD