一种基于Alphapose和LSTM的FMS打分模型
FMS scoring model based on alphapose and LSTM
近年来，随着健身爱好者的增加，功能性动作筛查（Functional Movement Screen, FMS）开始慢慢的进入大家的视野辅助大众进行健身训练，但是传统的FMS打分主要是由人工或者传感器进行打分，本文首先创建了FMS-深蹲的数据集，然后提出了一种基于Alphapose和LSTM神经网络的FMS深蹲打分模型。该模型通过Alphapose对连续多帧的测试者进行骨骼关键点检测，然后根据深蹲动作的打分标准对关键点坐标进行数据预处理，最后将数据输入到LSTM中进行训练。实验表明本文方法的准确率达到了89.6%。
In recent years, with the increase of fitness enthusiasts, functional motion screening (FMS) has gradually come into our view to assist the public in fitness training, but the traditional FMS scoring is mainly by artificial or sensor scoring. This paper first creates the FMS squat logarithm data set, and then proposes a FMS squat scoring model based on alphase and LSTM neural network. In this model, alphapose is used to detect the skeleton key points of the continuous multi frame testers, and then the data of the key points coordinates is preprocessed according to the scoring standard of squat movement, and finally the data is input into LSTM for training. The experimental results show that the accuracy of this method is 89.6%
罗志刚、游向东
体育计算技术、计算机技术
功能性动作筛查AlphaposeLSTM?????
FMSlphaposeLSTM
罗志刚,游向东.一种基于Alphapose和LSTM的FMS打分模型[EB/OL].(2021-03-16)[2025-08-11].http://www.paper.edu.cn/releasepaper/content/202103-156.点此复制
评论