Integrating Part-Object Relationship and Contrast for Camouflaged Object Detection

Object detectors that solely rely on image contrast are struggling to detect camouflaged objects in images because of the high similarity between and their surroundings. To address this issue, paper, we investigate role part-object relationship for object detection. Specifically, propose a Part-Object Contrast Integrated Network (POCINet) covering both search identification stages, where each stage adopts an appropriate scheme engage information relational knowledge pattern decoding. Besides, bridge these two stages via Search-to-Identification Guidance (SIG) module, which result, as well decoded semantic knowledge, jointly enhances features encoding ability stage. Experimental results demonstrate superiority our algorithm three datasets. Notably, raises F? best existing method by approximately 17 points CPD1K dataset. The source code will be released soon.