Fine-Grained Gesture Recognition Based on Machine Learning via RFID
Yuankang Wang, Yajun Zhang · 2024
Gesture recognition is becoming an essential part of smart living. Commercial RFID devices have played a crucial role in advancing gesture recognition due to their low cost, broad applicability, and ease of deployment. This paper proposes a commercial RFID-based contactless gesture recognition system that utilizes an improved Varri segmentation algorithm to accurately identify the start and end segments of gesture signals. By employing wavelet decomposition, fine-grained features of the gestures are extracted, enabling efficient capture of dynamic, detailed gesture signals. Evaluated through extensive gesture recognition experiments simulating real-world scenarios, this method achieves an average recognition accuracy of 98.2% in an empty room and 97.4% in a noisy classroom. The results demonstrate high recognition accuracy and robustness of the method.