{"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/can-adversarial-networks-hallucinate-occluded","title":"Can Adversarial Networks Hallucinate Occluded People With a Plausible Aspect?","arxiv_id":"1901.08097","date":"2019-01-23","proceeding":null,"authors":["Federico Fulgeri","Matteo Fabbri","Stefano Alletto","Simone Calderara","Rita Cucchiara"],"abstract":"When you see a person in a crowd, occluded by other persons, you miss visual\ninformation that can be used to recognize, re-identify or simply classify him\nor her. You can imagine its appearance given your experience, nothing more.\nSimilarly, AI solutions can try to hallucinate missing information with\nspecific deep learning architectures, suitably trained with people with and\nwithout occlusions. The goal of this work is to generate a complete image of a\nperson, given an occluded version in input, that should be a) without occlusion\nb) similar at pixel level to a completely visible people shape c) capable to\nconserve similar visual attributes (e.g. male/female) of the original one. For\nthe purpose, we propose a new approach by integrating the state-of-the-art of\nneural network architectures, namely U-nets and GANs, as well as discriminative\nattribute classification nets, with an architecture specifically designed to\nde-occlude people shapes. The network is trained to optimize a Loss function\nwhich could take into account the aforementioned objectives. As well we propose\ntwo datasets for testing our solution: the first one, occluded RAP, created\nautomatically by occluding real shapes of the RAP dataset (which collects also\nattributes of the people aspect); the second is a large synthetic dataset, AiC,\ngenerated in computer graphics with data extracted from the GTA video game,\nthat contains 3D data of occluded objects by construction. Results are\nimpressive and outperform any other previous proposal. This result could be an\ninitial step to many further researches to recognize people and their behavior\nin an open crowded world.","url_abs":"http://arxiv.org/abs/1901.08097v1","url_pdf":"http://arxiv.org/pdf/1901.08097v1.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":"can-adversarial-networks-hallucinate-occluded","repo_url":"https://github.com/fabbrimatteo/AiC-Dataset","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"attribute","task_name":"Attribute"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}