Work

Does SAM dream of EIG? Characterizing interactive segmenter performance using expected information gain

Year: 2025

Type: conference-paper

Abstract: We introduce an assessment procedure for interactive segmentation models. Based on concepts from Bayesian Experimental Design, the procedure measures a model’s understanding of point prompts and their correspondence with the desired segmentation mask. We demonstrate the use of the proposed procedure on out-of-the-box SAM and three medical datasets.

Source: SPIE Medical Imaging: Image Processing

DOI: 10.1117/12.3047159

Authors: Kuan-I Chung, Daniel Moyer

Language: English

Topic: Medical Image Segmentation Techniques

Subfield: Computer Vision and Pattern Recognition

Field: Computer Science

Domain: Physical Sciences

Open Access status: green

OpenAlex ID: W4407568874

FWCI: 0.783

Cites: 20

Cited by: 1

Related to: 10