{"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/totally-looks-like-how-humans-compare","title":"Totally Looks Like - How Humans Compare, Compared to Machines","arxiv_id":"1803.01485","date":"2018-03-05","proceeding":null,"authors":["Amir Rosenfeld","Markus D. Solbach","John K. Tsotsos"],"abstract":"Perceptual judgment of image similarity by humans relies on rich internal\nrepresentations ranging from low-level features to high-level concepts, scene\nproperties and even cultural associations. However, existing methods and\ndatasets attempting to explain perceived similarity use stimuli which arguably\ndo not cover the full breadth of factors that affect human similarity\njudgments, even those geared toward this goal. We introduce a new dataset\ndubbed Totally-Looks-Like (TLL) after a popular entertainment website, which\ncontains images paired by humans as being visually similar. The dataset\ncontains 6016 image-pairs from the wild, shedding light upon a rich and diverse\nset of criteria employed by human beings. We conduct experiments to try to\nreproduce the pairings via features extracted from state-of-the-art deep\nconvolutional neural networks, as well as additional human experiments to\nverify the consistency of the collected data. Though we create conditions to\nartificially make the matching task increasingly easier, we show that\nmachine-extracted representations perform very poorly in terms of reproducing\nthe matching selected by humans. We discuss and analyze these results,\nsuggesting future directions for improvement of learned image representations.","url_abs":"http://arxiv.org/abs/1803.01485v3","url_pdf":"http://arxiv.org/pdf/1803.01485v3.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":[],"tasks":[],"methods":[],"datasets_introduced":[{"slug":"tll","name":"TLL","full_name":"Totally-Looks-Like"}],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1803.01485","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}