Weighted Dense Semantic Aggregation and Explicit Boundary Modeling for Camouflaged Object Detection

Camouflaged object detection (COD) in monocular images has garnered broad attention recently, aiming to segment objects that have high intrinsic similarity with their surroundings. Despite remarkable performance achieved by existing methods, two limitations persist: insufficient utilization of multilevel semantics at each decoding scale and a lack of "explicit" knowledge guidance in boundary learning, leading to performance drops in challenging scenarios. To address these issues, we propose a weighted dense semantic aggregation (WDSA) and explicit boundary modeling (EBM) network. Specifically, a WDSA module is proposed to sufficiently aggregate multilevel semantics at each decoding scale, and enable the exploration of the relationship between multilevel features and camouflaged objects. An EBM module is developed to capture edge semantics with explicit boundary knowledge guidance and enhance the feature representation with edge cues. A detail enhanced multiscale (DEMS) module is further designed to refine multiscale features. Extensive experiments demonstrate that our proposed method achieves competitive performance against state-of-the-art (SOTA) methods on four benchmark datasets without excessive model complexity. Codes and results will be released at https://github.com/crrcoo/SAE-Net .