{"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/invariance-analysis-of-saliency-models-versus","title":"Invariance Analysis of Saliency Models versus Human Gaze During Scene Free Viewing","arxiv_id":"1810.04456","date":"2018-10-10","proceeding":null,"authors":["Zhaohui Che","Ali Borji","Guangtao Zhai","Xiongkuo Min"],"abstract":"Most of current studies on human gaze and saliency modeling have used\nhigh-quality stimuli. In real world, however, captured images undergo various\ntypes of distortions during the whole acquisition, transmission, and displaying\nchain. Some distortion types include motion blur, lighting variations and\nrotation. Despite few efforts, influences of ubiquitous distortions on visual\nattention and saliency models have not been systematically investigated. In\nthis paper, we first create a large-scale database including eye movements of\n10 observers over 1900 images degraded by 19 types of distortions. Second, by\nanalyzing eye movements and saliency models, we find that: a) observers look at\ndifferent locations over distorted versus original images, and b) performances\nof saliency models are drastically hindered over distorted images, with the\nmaximum performance drop belonging to Rotation and Shearing distortions.\nFinally, we investigate the effectiveness of different distortions when serving\nas data augmentation transformations. Experimental results verify that some\nuseful data augmentation transformations which preserve human gaze of reference\nimages can improve deep saliency models against distortions, while some invalid\ntransformations which severely change human gaze will degrade the performance.","url_abs":"http://arxiv.org/abs/1810.04456v1","url_pdf":"http://arxiv.org/pdf/1810.04456v1.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":"invariance-analysis-of-saliency-models-versus","repo_url":"https://github.com/CZHQuality/Sal-CFS-GAN","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"data-augmentation","task_name":"Data Augmentation"}],"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}