Deep learning-based 2D keypoint detection in alpine ski racing – A performance analysis of state-of-the-art algorithms applied to regular skiing and injury situations
Michael Zwölfer , Dieter Heinrich , Kurt Schindelwig , Bastian Wandt , Helge Rhodin , Jörg Spörri , Werner Nachbauer
求助PDF
{"title":"Deep learning-based 2D keypoint detection in alpine ski racing – A performance analysis of state-of-the-art algorithms applied to regular skiing and injury situations","authors":"Michael Zwölfer , Dieter Heinrich , Kurt Schindelwig , Bastian Wandt , Helge Rhodin , Jörg Spörri , Werner Nachbauer","doi":"10.1016/j.jsampl.2023.100034","DOIUrl":null,"url":null,"abstract":"","PeriodicalId":74029,"journal":{"name":"JSAMS plus","volume":"2 ","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"JSAMS plus","FirstCategoryId":"1085","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2772696723000157","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 1
引用
批量引用
高山滑雪比赛中基于深度学习的2D关键点检测-应用于常规滑雪和受伤情况的最先进算法的性能分析
目的在本研究中,我们检验了基于深度学习的2D关键点检测应用于高山滑雪赛道上常规滑雪和受伤情况(即失去平衡和摔倒情况)的实用性。因此,我们创建了一个常规的滑雪和受伤情况特定数据集(以下简称“受伤滑雪数据集”),在该数据集上比较了最先进的关键点检测算法OpenPose、Mask-R-NN、AlphaPose和DCPose。通过计算每个关节位置误差的平均值(MPJPE)和正确关键点的百分比(PCK)来评估每个关键点检测器的性能。通过可视化分析进一步调查了故障案例和常见错误模式。结果我们观察到常规滑雪的最佳结果,在9(2)到14(3)像素的MPJPE下,81%–92%的关键点被正确检测到。在受伤的情况下,自我遮挡和罕见姿势变得更有可能,类似于雪雾和运动模糊造成的遮挡。因此,在失衡情况下,性能下降到68%-80%(PCK),而在秋季情况下,只有35%-54%的关键点被正确检测到,平均误差为26-36像素。在所有算法中,AlphaPose是最稳健的,并取得了最好的结果。结论常规滑雪的sPCK和MPJPE在手动注释错误的范围内,可以认为其足够低,可以进行进一步的生物力学分析。对于跌倒情况,应进一步改进关键点检测。关于未来开发用于高山滑雪损伤分析的深度学习工具,我们建议在滑雪和损伤特定数据集(如我们的数据集)上微调性能良好的关键点检测器,如AlphaPose。
本文章由计算机程序翻译,如有差异,请以英文原文为准。