Cosalpure: Learning Concept from Group Images for Robust Co-Saliency Detection

Concept Learning Concept-guided Purification Co-salient Object Detectors c T2I Diffusion Group Images Containing some Adv. Examples Purified Group Images Detection Results Detection Results CosalPure Validation of c Learned concept Figure 1 . Examples of our method COSALPURE and comparative results before and after purification. COSALPURE comprises two modules: group-image concept learning and concept-guided purification. Firstly, the concept learning module inputs a group of images that contain some adversarial cases and obtain their shared co-salient semantic information (i.e., the learned concept), denoted as c. We can validate the effectiveness of the learned c through the visualization via a text-to-image (T2I) diffusion model. Secondly, steered by the previously learned concept, we employ certain diffusion generation techniques to purify the entire group of images. Before our purification, the co-salient object detection results are poor, but after purification, the detection results are satisfactory. Please enlarge to see more details.