MIAdapt: Source-free Few-shot Domain Adaptive Object Detection for Microscopic Images

2025-03-05

AI Summary

This paper addresses the challenging problem of domain adaptive object detection in medical microscopic imaging, where source-domain data are unavailable and only a few labeled samples exist in the target domain. To this end, we formally define and introduce a novel task: Source-Free Few-Shot Domain Adaptive Detection (SF-FSDA). We propose two dual-baseline methods—Faster-FreeShot and MT-FreeShot—constituting the first source-free few-shot adaptive detection framework that operates without access to any source-domain images during training. Our approach integrates model distillation, consistency regularization, and dynamic pseudo-label optimization. Evaluated on the Raabin-WBC dataset, our methods achieve mAP improvements of 21.3% over existing source-free unsupervised domain adaptation (UDA) methods and 4.7% over state-of-the-art few-shot domain adaptation (DA) approaches. These results demonstrate significantly enhanced generalization under extreme label scarcity, establishing a new paradigm for privacy-preserving and data-efficient analysis of medical imagery.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Computer Vision: Medical and Biological Imaging
• Application Domains: Misinformation & Fake News

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Web Mining and Content Analysis: Large pretrained models with web data
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Existing generic unsupervised domain adaptation approaches require access to both a large labeled source dataset and a sufficient unlabeled target dataset during adaptation. However, collecting a large dataset, even if unlabeled, is a challenging and expensive endeavor, especially in medical imaging. In addition, constraints such as privacy issues can result in cases where source data is unavailable. Taking in consideration these challenges, we propose MIAdapt, an adaptive approach for Microscopic Imagery Adaptation as a solution for Source-free Few-shot Domain Adaptive Object detection (SF-FSDA). We also define two competitive baselines (1) Faster-FreeShot and (2) MT-FreeShot. Extensive experiments on the challenging M5-Malaria and Raabin-WBC datasets validate the effectiveness of MIAdapt. Without using any image from the source domain MIAdapt surpasses state-of-the-art source-free UDA (SF-UDA) methods by +21.3% mAP and few-shot domain adaptation (FSDA) approaches by +4.7% mAP on Raabin-WBC. Our code and models will be publicly available.

Problem

• Addresses source-free few-shot domain adaptation in microscopic images.
• Proposes MIAdapt for object detection without source domain images.
• Validates MIAdapt on M5-Malaria and Raabin-WBC datasets effectively.

Innovation

• Source-free domain adaptation for microscopic images
• Few-shot learning without source domain images
• Competitive baselines: Faster-FreeShot, MT-FreeShot