{"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/data-distillation-towards-omni-supervised","title":"Data Distillation: Towards Omni-Supervised Learning","arxiv_id":"1712.04440","date":"2017-12-12","proceeding":"CVPR 2018 6","authors":["Ilija Radosavovic","Piotr Dollár","Ross Girshick","Georgia Gkioxari","Kaiming He"],"abstract":"We investigate omni-supervised learning, a special regime of semi-supervised\nlearning in which the learner exploits all available labeled data plus\ninternet-scale sources of unlabeled data. Omni-supervised learning is\nlower-bounded by performance on existing labeled datasets, offering the\npotential to surpass state-of-the-art fully supervised methods. To exploit the\nomni-supervised setting, we propose data distillation, a method that ensembles\npredictions from multiple transformations of unlabeled data, using a single\nmodel, to automatically generate new training annotations. We argue that visual\nrecognition models have recently become accurate enough that it is now possible\nto apply classic ideas about self-training to challenging real-world data. Our\nexperimental results show that in the cases of human keypoint detection and\ngeneral object detection, state-of-the-art models trained with data\ndistillation surpass the performance of using labeled data from the COCO\ndataset alone.","url_abs":"http://arxiv.org/abs/1712.04440v1","url_pdf":"http://arxiv.org/pdf/1712.04440v1.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":"data-distillation-towards-omni-supervised","repo_url":"https://github.com/EliPassov/classification-ensembles","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"data-distillation-towards-omni-supervised","repo_url":"https://github.com/facebookresearch/detectron","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"data-distillation-towards-omni-supervised","repo_url":"https://github.com/jiajunhua/facebookresearch-Detectron","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"caffe2","reach":null},{"paper_slug":"data-distillation-towards-omni-supervised","repo_url":"https://github.com/notha99y/mean_teacher_domain_adapt","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"keypoint-detection","task_name":"Keypoint Detection"},{"task_slug":"object-detection","task_name":"Object Detection"},{"task_slug":"object-detection-1","task_name":"object-detection"}],"methods":[{"method_slug":"1x1-convolution","method_name":"1x1 Convolution"},{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"fpn","method_name":"FPN"},{"method_slug":"faster-r-cnn","method_name":"Faster R-CNN"},{"method_slug":"mask-r-cnn","method_name":"Mask R-CNN"},{"method_slug":"rpn","method_name":"RPN"},{"method_slug":"roi-align","method_name":"RoIAlign"},{"method_slug":"roipool","method_name":"RoIPool"},{"method_slug":"softmax","method_name":"Softmax"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1712.04440","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}