Red-Teaming Segment Anything Model

arXiv:2404.02067v1 Announce Type: cross
Abstract: Foundation models have emerged as pivotal tools, tackling many complex tasks through pre-training on vast datasets and subsequent fine-tuning for specific applications. The Segment Anything Model is one of the first and most well-known foundation models for computer vision segmentation tasks. This work presents a multi-faceted red-teaming analysis that tests the Segment Anything Model against challenging tasks: (1) We analyze the impact of style transfer on segmentation masks, demonstrating that applying adverse weather conditions and …

abstract analysis applications arxiv computer computer vision cs.ai cs.cv cs.lg datasets fine-tuning foundation pivotal pre-training segment segment anything segment anything model segmentation tasks tests through tools training type vast vision work