Classifying weight training workouts with deep convolutional neural networks: a precedent study

Abstract

In recent years, deep learning algorithms have been widely used in both academic research and practical applications. This study uses a deep convolutional neural network to analyze and predict physical movements. We evaluated the effectiveness of our proposed network by recruiting a professional fitness trainer and let the trainer wear a smart watch equipped with an accelerometer capable of assessing physical movement. The results confirmed the ability of the network to correctly predict the bench press, dips, squat, deadlift, and military press with an accuracy rate of 92.8%. This preliminary study has several limitations such as a low sample size and the lack of a specified network layer. In subsequent studies we plan to address these limitations by extending our investigation to include the analysis of diverse movements.