{"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/in-bed-pose-estimation-deep-learning-with","title":"In-Bed Pose Estimation: Deep Learning with Shallow Dataset","arxiv_id":"1711.01005","date":"2017-11-03","proceeding":null,"authors":["Shuangjun Liu","Yu Yin","Sarah Ostadabbas"],"abstract":"Although human pose estimation for various computer vision (CV) applications\nhas been studied extensively in the last few decades, yet in-bed pose\nestimation using camera-based vision methods has been ignored by the CV\ncommunity because it is assumed to be identical to the general purpose pose\nestimation methods. However, in-bed pose estimation has its own specialized\naspects and comes with specific challenges including the notable differences in\nlighting conditions throughout a day and also having different pose\ndistribution from the common human surveillance viewpoint. In this paper, we\ndemonstrate that these challenges significantly lessen the effectiveness of\nexisting general purpose pose estimation models. In order to address the\nlighting variation challenge, infrared selective (IRS) image acquisition\ntechnique is proposed to provide uniform quality data under various lighting\nconditions. In addition, to deal with unconventional pose perspective, a 2-end\nhistogram of oriented gradient (HOG) rectification method is presented. In this\nwork, we explored the idea of employing a pre-trained convolutional neural\nnetwork (CNN) model trained on large public datasets of general human poses and\nfine-tuning the model using our own shallow in-bed IRS dataset. We developed an\nIRS imaging system and collected IRS image data from several realistic\nlife-size mannequins in a simulated hospital room environment. A pre-trained\nCNN called convolutional pose machine (CPM) was repurposed for in-bed pose\nestimation by fine-tuning its specific intermediate layers. Using the HOG\nrectification method, the pose estimation performance of CPM significantly\nimproved by 26.4% in PCK0.1 criteria compared to the model without such\nrectification.","url_abs":"http://arxiv.org/abs/1711.01005v3","url_pdf":"http://arxiv.org/pdf/1711.01005v3.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":"in-bed-pose-estimation-deep-learning-with","repo_url":"https://github.com/ostadabbas/in-bed-pose-estimation","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"BSD-3-Clause"}}],"tasks":[{"task_slug":"deep-learning","task_name":"Deep Learning"},{"task_slug":"pose-estimation","task_name":"Pose Estimation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}