From IMUs to Smartwatches: Measuring Performance in Practical Shooting

Human Activity Recognition (HAR) has seen significant advancements through wearable sensors and machine learning, particularly in sports like running and swimming. However, high-speed tactical disciplines such as practical shooting remain underexplored due to their complex, overlapping actions and the need for non-intrusive equipment. This study investigates the feasibility of using commercial smartwatches and smartphones to replace laboratory-grade inertial measurement units (IMUs) for performance monitoring in practical shooting. Twelve athletes participated in trials, with data collected using full-body IMUs and reduced sensor setups. Key actions - shooting, running, reloading, and extraction - were identified and classified using binary classifiers and multi-label strategies. Results showed that a combination of sensors on the dominant wrist and hip provided optimal accuracy, outperforming single-sensor setups and audio-based configurations. Additionally, training classifiers on complete exercise sessions yielded better performance than isolated actions. The findings demonstrate the potential of commercial devices for HAR in practical shooting, offering a cost-effective and energy-efficient solution for real-world deployment. This approach could be extended to other sports, balancing accuracy, cost, and usability.