Unsupervised Camouflaged Object Segmentation as Domain Adaptation

Deep learning for unsupervised image segmentation re- mains challenging due to the absence of human labels. The common idea is to train a segmentation head, with the su- pervision of pixel-wise pseudo-labels generated based on the representation of self-supervised backbones. By doing so, the model performance depends much on the distance between the distribution of target datasets, and the one of backbones’ pre-training dataset (e.g., ImageNet). In this work, we investigate a new task, namely unsupervised cam- ouflaged object segmentation (UCOS), where the target ob- jects own a common rarely-seen attribute, i.e., camouflage. Unsurprisingly, we find that the state-of-the-art unsuper- vised models struggle in adapting UCOS, due to the domain gap between the properties of generic and camouflaged ob- jects. To this end, we formulate the UCOS as a source-free unsupervised domain adaptation task (UCOS-DA), where both source labels and target labels are absent during the whole model training process. Specifically, we define a source model consisting of self-supervised vision transform- ers pre-trained on ImageNet. On the other hand, the tar- get domain includes a simple linear layer (i.e., our target model) and unlabeled camouflaged objects. We then de- sign a pipeline for foreground-background-contrastive self- adversarial domain adaptation, to achieve robust UCOS. As a result, our baseline model achieves superior segmenta- tion performance when compared with competing unsuper- vised models on the UCOS benchmark, with the training set which’s scale is only one tenth of the supervised COS coun- terpart. The UCOS benchmark and our baseline model are now publicly available