SAM Struggles in Concealed Scenes -- Empirical Study on Segment Anything

TL;DR

This paper empirically evaluates the Segment Anything Model (SAM) in concealed scenes like camouflaged animals, industrial defects, and medical lesions, revealing its limitations in unprompted settings for such challenging scenarios.

Contribution

It provides the first empirical analysis of SAM's performance in concealed scenes, highlighting its struggles and limitations in these complex contexts.

Findings

SAM performs poorly in concealed scenes

SAM struggles with unprompted segmentation tasks

Highlights need for improved models in challenging scenarios

Abstract

Segmenting anything is a ground-breaking step toward artificial general intelligence, and the Segment Anything Model (SAM) greatly fosters the foundation models for computer vision. We could not be more excited to probe the performance traits of SAM. In particular, exploring situations in which SAM does not perform well is interesting. In this report, we choose three concealed scenes, i.e., camouflaged animals, industrial defects, and medical lesions, to evaluate SAM under unprompted settings. Our main observation is that SAM looks unskilled in concealed scenes.