Gait-Recognition

This repository contains all files related to the project I had done on Human Identification using Gait Recognition. I used dataset obtained from CASIA-B Gait Database and pre-processed it to get normalised and de-noised binary silhouettes. Then I creatted Pose Energy Image as the key feature for classification in my project, according to which a gait cycle is divided into a number of key poses. The binary silhouettes obtained from registered frames corresponding to each key pose were classified using Dijkstra’s Shortest Path Algorithm to its respective Pose and were concatenated to generate a final feature. Classification was then done using KNN Classifier.