Gym Exercises Monitoring with Smart Gloves: Exercise Recognition, Repetition Counting, and Imbalance Quantification
Muhammad Waleed Aslam, Zhen Liang, XU Guang-hua, Wenwu Deng, Qijun Ying, Jingyuan Cheng · 2023
Fitness training attracts a lot of attention, as it tumbles the health risks in a participant. Most fitness training activities involve the interaction of hand and workout equipment. Therefore, we present a pressure-sensing smart glove system to recognize fitness exercises, count repetitions, and quantify imbalance. The proposed smart glove with 93 sensing points at the wrist and palm can attain pressure-distributed time series data. We evaluate the system’s performance using the collected exercise data from 20 participants. Support vector classifier attains overall 96 accuracy of exercise recognition. For counting repetitions, the overall mean absolute error of 1.78 is calculated for ten types of exercises. Using the ratio index technique, we compute the upper-threshold (+0.6) and lower-threshold (-0.6) to quantify the imbalance of lifted weights.