Sign Language Recognition using Gait Energy Image

Gait Recognition with 3D CNN. This project proposes a novel approach using 3D convolutional neural networks (3D CNN) to capture spatio-temporal features of gait sequences for robust recognition in an un-intrusive manner.

This project leverages state-of-the-art deep learning models—VGG-16, InceptionV1, ViT, EfficientNet-B0, and ResNet50—for gait image classification. By applying decision level fusion, feature level fusion, and hybrid fusion techniques, the project achieves enhanced accuracy, making it suitable for any image classification task.