CamoSAM2: SAM2-oriented Prompt Auto-Refinement for Video Camouflaged Object Detection
Researchers have developed CamoSAM2, a new framework designed to improve video camouflaged object detection (VCOD) using the SAM2 foundation model. This system automatically generates and refines prompts for SAM2, addressing challenges in perceiving and reliably prompting camouflaged objects. CamoSAM2 integrates motion and appearance cues for initial predictions and employs a video-based adaptive multi-prompts refinement strategy, achieving significant improvements in mean Intersection over Union (mIoU) and faster inference speeds compared to existing methods. AI
IMPACT Improves automated detection of camouflaged objects in videos, potentially aiding applications in surveillance and environmental monitoring.