{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/picanet-learning-pixel-wise-contextual","title":"PiCANet: Learning Pixel-wise Contextual Attention for Saliency Detection","arxiv_id":"1708.06433","date":"2017-08-21","proceeding":"CVPR 2018 6","authors":["Nian Liu","Junwei Han","Ming-Hsuan Yang"],"abstract":"Contexts play an important role in the saliency detection task. However,\ngiven a context region, not all contextual information is helpful for the final\ntask. In this paper, we propose a novel pixel-wise contextual attention\nnetwork, i.e., the PiCANet, to learn to selectively attend to informative\ncontext locations for each pixel. Specifically, for each pixel, it can generate\nan attention map in which each attention weight corresponds to the contextual\nrelevance at each context location. An attended contextual feature can then be\nconstructed by selectively aggregating the contextual information. We formulate\nthe proposed PiCANet in both global and local forms to attend to global and\nlocal contexts, respectively. Both models are fully differentiable and can be\nembedded into CNNs for joint training. We also incorporate the proposed models\nwith the U-Net architecture to detect salient objects. Extensive experiments\nshow that the proposed PiCANets can consistently improve saliency detection\nperformance. The global and local PiCANets facilitate learning global contrast\nand homogeneousness, respectively. As a result, our saliency model can detect\nsalient objects more accurately and uniformly, thus performing favorably\nagainst the state-of-the-art methods.","url_abs":"http://arxiv.org/abs/1708.06433v2","url_pdf":"http://arxiv.org/pdf/1708.06433v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"picanet-learning-pixel-wise-contextual","repo_url":"https://github.com/nian-liu/PiCANet","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"none","reach":null},{"paper_slug":"picanet-learning-pixel-wise-contextual","repo_url":"https://github.com/Ugness/PiCANet-Implementation","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"salient-object-detection","task_name":"RGB Salient Object Detection"},{"task_slug":"saliency-detection","task_name":"Saliency Detection"}],"methods":[{"method_slug":"concatenated-skip-connection","method_name":"Concatenated Skip Connection"},{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"max-pooling","method_name":"Max Pooling"},{"method_slug":"relu","method_name":"ReLU"},{"method_slug":"u-net","method_name":"U-Net"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/salient-object-detection-on-duts-te","task":"RGB Salient Object Detection","dataset":"DUTS-TE","model":"PiCANet","rank_in_archive_order":15,"of":31,"metrics":{"MAE":"0.050","S-Measure":"0.842","max F-measure":"0.863","mean E-Measure":"0.853","mean F-Measure":"0.757"},"uses_additional_data":false},{"leaderboard":"/sota/salient-object-detection-on-soc","task":"RGB Salient Object Detection","dataset":"SOC","model":"PiCANet","rank_in_archive_order":7,"of":7,"metrics":{"Average MAE":"0.133","S-Measure":"0.801","mean E-Measure":"0.810"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}