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

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.

Accepted by SCIENCE CHINA Information Sciences, 2023

References in corpus (2)

• Anabranch Network for Camouflaged Object Segmentation
• CamoFormer: Masked Separable Attention for Camouflaged Object Detection

Cited by in corpus (10)

• Segment Anything in Medical Images
• Segment Anything Model for Medical Images?
• Segment Anything Is Not Always Perfect: An Investigation of SAM on Different Real-world Applications
• Generalist Vision Foundation Models for Medical Imaging: A Case Study of Segment Anything Model on Zero-Shot Medical Segmentation
• Advances in Deep Concealed Scene Understanding
• Towards Real Zero-Shot Camouflaged Object Segmentation without Camouflaged Annotations
• Don't waste SAM
• 3D Object Detection and High-Resolution Traffic Parameters Extraction Using Low-Resolution LiDAR Data
• Weakly Supervised Camouflaged Object Detection Based on the SAM Model and Mask Guidance
• Self-Supervised Real-Time Tracking of Military Vehicles in Low-FPS UAV Footage