Description of Invention
This project explores a novel deep learning-based approach for biometric identification through human gait, offering a contactless and unobtrusive alternative to traditional methods like fingerprint or facial recognition. Gait recognition capitalizes on the distinctive way each individual walks. The challenge lies in achieving consistent accuracy across varying clothing styles, occlusions, and lighting. This research aims to develop a robust, high-performing system using Convolutional Neural Networks (CNNs), specifically leveraging InceptionResNetV2, a hybrid of Inception modules and Residual Networks, which achieved a remarkable 99.78% test accuracy.