{"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/mpiigaze-real-world-dataset-and-deep","title":"MPIIGaze: Real-World Dataset and Deep Appearance-Based Gaze Estimation","arxiv_id":"1711.09017","date":"2017-11-24","proceeding":null,"authors":["Xucong Zhang","Yusuke Sugano","Mario Fritz","Andreas Bulling"],"abstract":"Learning-based methods are believed to work well for unconstrained gaze\nestimation, i.e. gaze estimation from a monocular RGB camera without\nassumptions regarding user, environment, or camera. However, current gaze\ndatasets were collected under laboratory conditions and methods were not\nevaluated across multiple datasets. Our work makes three contributions towards\naddressing these limitations. First, we present the MPIIGaze that contains\n213,659 full face images and corresponding ground-truth gaze positions\ncollected from 15 users during everyday laptop use over several months. An\nexperience sampling approach ensured continuous gaze and head poses and\nrealistic variation in eye appearance and illumination. To facilitate\ncross-dataset evaluations, 37,667 images were manually annotated with eye\ncorners, mouth corners, and pupil centres. Second, we present an extensive\nevaluation of state-of-the-art gaze estimation methods on three current\ndatasets, including MPIIGaze. We study key challenges including target gaze\nrange, illumination conditions, and facial appearance variation. We show that\nimage resolution and the use of both eyes affect gaze estimation performance\nwhile head pose and pupil centre information are less informative. Finally, we\npropose GazeNet, the first deep appearance-based gaze estimation method.\nGazeNet improves the state of the art by 22% percent (from a mean error of 13.9\ndegrees to 10.8 degrees) for the most challenging cross-dataset evaluation.","url_abs":"http://arxiv.org/abs/1711.09017v1","url_pdf":"http://arxiv.org/pdf/1711.09017v1.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":"mpiigaze-real-world-dataset-and-deep","repo_url":"https://github.com/dreamer-1996/eye-tracking-with-resnet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"mpiigaze-real-world-dataset-and-deep","repo_url":"https://github.com/hysts/pytorch_mpiigaze","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"mpiigaze-real-world-dataset-and-deep","repo_url":"https://github.com/kenkyusha/eyeGazeToScreen","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"mpiigaze-real-world-dataset-and-deep","repo_url":"https://github.com/kroniidvul/mpiigaze_project","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"mpiigaze-real-world-dataset-and-deep","repo_url":"https://github.com/songNew/3_Gaze_MPII","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"mpiigaze-real-world-dataset-and-deep","repo_url":"https://github.com/trakaros/MPIIGaze","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"none","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"gaze-estimation","task_name":"Gaze Estimation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1711.09017","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}