Wi-Finish: Exploring Wi-Fi Sensing for Detecting Athletic Finish Line Crossings on Resource-Constrained Platforms

Athletic timing systems are essential for fair competition and performance assessment, yet traditional solutions are often expensive and cumbersome to deploy. This paper explores the application of Wi-Fi sensing to detect runners crossing a finish line using resource-constrained platforms. We propose the use of lightweight machine learning models, specifically Support Vector Machines (SVMs), as an alternative to deep learning, enabling data collection, processing, and classification on a single ESP32 microprocessor. The system collects and analyses Channel State Information (CSI) data to identify crossing events with high accuracy. Experimental results show that the SVM achieves an F1 score of 0.98 while maintaining a memory footprint of 8.4 KB, making it suitable for embedded systems. This study demonstrates the feasibility of low-cost timing systems for athletic applications based on Wi-Fi sensing.