RPG-SAM: Reliability-Weighted Prototypes and Geometric Adaptive Threshold Selection for Training-Free One-Shot Polyp Segmentation

2026-03-08

AI Summary

This work addresses a critical limitation in existing training-free one-shot polyp segmentation methods, which overlook the regional heterogeneity in support images and the intensity heterogeneity in query responses, thereby constraining segmentation accuracy. To overcome this, we propose RPG-SAM, a novel framework that systematically models both types of heterogeneity for the first time. Our approach integrates reliability-weighted prototype mining (RWPM), background-anchor contrastive suppression, geometrically adaptive threshold selection (GAS), and an iterative boundary refinement mechanism to achieve high-precision segmentation without any training. Evaluated on the Kvasir dataset, RPG-SAM achieves a 5.56% absolute improvement in mIoU over current training-free methods, demonstrating its superior performance and effectiveness.

Technology Category

• Computer Vision: Segmentation
• Search and Optimization: Learning to Search
• Machine Learning: Multi-instance/Multi-view Learning

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Security and Privacy: Data transparency and provenance
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Training-free one-shot segmentation offers a scalable alternative to expert annotations where knowledge is often transferred from support images and foundation models. But existing methods often treat all pixels in support images and query response intensities models in a homogeneous way. They ignore the regional heterogeity in support images and response heterogeity in query.To resolve this, we propose RPG-SAM, a framework that systematically tackles these heterogeneity gaps. Specifically, to address regional heterogeneity, we introduce Reliability-Weighted Prototype Mining (RWPM) to prioritize high-fidelity support features while utilizing background anchors as contrastive references for noise suppression. To address response heterogeneity, we develop Geometric Adaptive Selection (GAS) to dynamically recalibrate binarization thresholds by evaluating the morphological consensus of candidates. Finally, an iterative refinement loop method is designed to polishes anatomical boundaries. By accounting for multi-layered information heterogeneity, RPG-SAM achieves a 5.56\% mIoU improvement on the Kvasir dataset. Code will be released.

Problem

• one-shot segmentation
• regional heterogeneity
• response heterogeneity
• polyp segmentation
• training-free

Innovation

• Reliability-Weighted Prototype Mining
• Geometric Adaptive Threshold Selection
• Training-Free One-Shot Segmentation
• Response Heterogeneity
• Iterative Refinement