REPETITIVE ACTION COUNTING THROUGH JOINT ANGLE ANALYSIS AND VIDEO TRANSFORMER TECHNIQUES

The quantification of repetitive movements, known as repetitive action counting, is critical in various applications, such as fitness tracking, rehabilitation, and manufacturing operation monitoring. Traditional methods predominantly relied on the estimation of red-green-and-blue (RGB) frames and body pose landmarks to identify the number of action repetitions. However, these methods suffer from several issues, such as instability under varying camera viewpoints, propensity for over-counting or under-counting, challenges in differentiating sub-actions, and inaccuracies in recognizing salient action poses, etc. Our method integrates joint angles with body pose landmarks to address these issues, thereby surpassing the performance benchmarks of existing state-of-the-art repetitive action counting methodologies. The efficacy of our approach is underscored by a Mean Absolute Error (MAE) of 0.211 and an Off-By-One Accuracy (OBOA) of 0.599 on a public repetitive action counting data set, RepCount [1]. Comprehensive experimental results demonstrate the effectiveness and robustness of our method.