A Novel and Robust Gait Recognition method based on Hybrid Learning Methodology

Abstract

Model-free methods for recognizing human gait rely on tracking the form and velocity of the person in motion. Recognizability at far distances with suitably low-resolution photos is a strength of this method. Extracting gait characteristics from gait frames using this method is a breeze. This model-free method can be approached from several angles. This paper's goal is to develop a novel approach to gait identification, dubbed a “Hybrid Learning Classifier,” that combines an AI technique with learning principles (HLC). First, a binary outline picture of a walking human is recognized from each frame using this approach. Second, an image processing technique is used to extract features from each individual frame. Important characteristics discussed here are stature, hand and leg length, and left-right and right-left distances. Finally, HLC is put to use in the form of evaluation and practice. We've done away with the need to use a reductant training vector when choosing a model to train on or adjusting any of the many parameters associated with doing so. All of our research here utilizes our gait database. Depending on which datasets are used for training and which for testing, different conclusions might be drawn. This paper's concluding portion offers appropriate verification of everything said, presented graphically and with precise description.