Dual Distillation for Few-Shot Anomaly Detection

2026-03-02

AI Summary

This work addresses the limitations of existing few-shot medical anomaly detection methods, which heavily rely on abundant normal data and exhibit poor generalization across anatomical structures. To overcome these challenges, the authors propose D²4FAD, a novel framework built upon a pre-trained encoder–student decoder architecture. D²4FAD introduces a dual distillation mechanism that uniquely combines knowledge distillation from query images with self-distillation from support images, complemented by a dynamic weighting strategy to assess the relevance of each support sample to the current query. Evaluated on a comprehensive dataset comprising 13,084 images across four organs, four imaging modalities, and five disease types, D²4FAD achieves state-of-the-art performance in few-shot medical anomaly detection, significantly enhancing model generalization under limited normal training samples.

Technology Category

• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Medical and Biological Imaging

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Anomaly detection is a critical task in computer vision with profound implications for medical imaging, where identifying pathologies early can directly impact patient outcomes. While recent unsupervised anomaly detection approaches show promise, they require substantial normal training data and struggle to generalize across anatomical contexts. We introduce D$^2$4FAD, a novel dual distillation framework for few-shot anomaly detection that identifies anomalies in previously unseen tasks using only a small number of normal reference images. Our approach leverages a pre-trained encoder as a teacher network to extract multi-scale features from both support and query images, while a student decoder learns to distill knowledge from the teacher on query images and self-distill on support images. We further propose a learn-to-weight mechanism that dynamically assesses the reference value of each support image conditioned on the query, optimizing anomaly detection performance. To evaluate our method, we curate a comprehensive benchmark dataset comprising 13,084 images across four organs, four imaging modalities, and five disease categories. Extensive experiments demonstrate that D$^2$4FAD significantly outperforms existing approaches, establishing a new state-of-the-art in few-shot medical anomaly detection. Code is available at https://github.com/ttttqz/D24FAD.

Problem

• few-shot anomaly detection
• medical imaging
• anomaly detection
• generalization
• limited normal data

Innovation

• dual distillation
• few-shot anomaly detection
• knowledge distillation
• learn-to-weight mechanism
• medical image analysis