TinyML-Based Gait Recognition for User Detection on an STM32L4 IoT Node

Gait recognition stands out as a robust biometric tool for discerning users. Yet, adapting it to devices with limited resources, such as the STM32L4 microcontroller, demands cutting-edge strategies. This research unveils an innovative TinyML methodology, leveraging a specially-designed Convolutional Neural Network (CNN). Notably, our bespoke CNN model showcased a remarkable 97.3% accuracy in pinpointing users on the STM32L4 system.