CamouFinder: Finding Camouflaged Instances in Images

In this paper, we investigate the interesting yet challenging problem of camouflaged instance segmentation. To end, first annotate available CAMO dataset at level. We also embed data augmentation in order to increase number training samples. Then, train different state-of-the-art segmentation on CAMO-instance data. Last but not least, develop an interactive user interface which demonstrates performance methods task The users are able compare results given input images. Our work is expected push envelope camouflage analysis problem.