{"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/a-semi-supervised-data-augmentation-approach","title":"A Semi-Supervised Data Augmentation Approach using 3D Graphical Engines","arxiv_id":"1808.02595","date":"2018-08-08","proceeding":null,"authors":["Shuangjun Liu","Sarah Ostadabbas"],"abstract":"Deep learning approaches have been rapidly adopted across a wide range of\nfields because of their accuracy and flexibility, but require large labeled\ntraining datasets. This presents a fundamental problem for applications with\nlimited, expensive, or private data (i.e. small data), such as human pose and\nbehavior estimation/tracking which could be highly personalized. In this paper,\nwe present a semi-supervised data augmentation approach that can synthesize\nlarge scale labeled training datasets using 3D graphical engines based on a\nphysically-valid low dimensional pose descriptor. To evaluate the performance\nof our synthesized datasets in training deep learning-based models, we\ngenerated a large synthetic human pose dataset, called ScanAva using 3D scans\nof only 7 individuals based on our proposed augmentation approach. A\nstate-of-the-art human pose estimation deep learning model then was trained\nfrom scratch using our ScanAva dataset and could achieve the pose estimation\naccuracy of 91.2% at PCK0.5 criteria after applying an efficient domain\nadaptation on the synthetic images, in which its pose estimation accuracy was\ncomparable to the same model trained on large scale pose data from real humans\nsuch as MPII dataset and much higher than the model trained on other synthetic\nhuman dataset such as SURREAL.","url_abs":"http://arxiv.org/abs/1808.02595v2","url_pdf":"http://arxiv.org/pdf/1808.02595v2.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":"a-semi-supervised-data-augmentation-approach","repo_url":"https://github.com/ostadabbas/ScanAvaGenerationToolkit","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"data-augmentation","task_name":"Data Augmentation"},{"task_slug":"deep-learning","task_name":"Deep Learning"},{"task_slug":"domain-adaptation","task_name":"Domain Adaptation"},{"task_slug":"pose-estimation","task_name":"Pose Estimation"},{"task_slug":"small-data","task_name":"Small Data Image Classification"},{"task_slug":null,"task_name":"valid"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}