Camouflaged Object Segmentation with Distraction Mining

Camouflaged object segmentation (COS) aims to identify objects that are"perfectly" assimilate into their surroundings, which has a wide range ofvaluable applications. The key challenge of COS is that there exist highintrinsic similarities between the candidate objects and noise background. Inthis paper, we strive to embrace challenges towards effective and efficientCOS. To this end, we develop a bio-inspired framework, termed Positioning andFocus Network (PFNet), which mimics the process of predation in nature.Specifically, our PFNet contains two key modules, i.e., the positioning module(PM) and the focus module (FM). The PM is designed to mimic the detectionprocess in predation for positioning the potential target objects from a globalperspective and the FM is then used to perform the identification process inpredation for progressively refining the coarse prediction via focusing on theambiguous regions. Notably, in the FM, we develop a novel distraction miningstrategy for distraction discovery and removal, to benefit the performance ofestimation. Extensive experiments demonstrate that our PFNet runs in real-time(72 FPS) and significantly outperforms 18 cutting-edge models on threechallenging datasets under four standard metrics.