For the purpose of automated face recognition, a method is suggested that combines gait, a behavioural biometric, with the face, a physical biometric. The unknown person's identity is determined using a decision-level fusion technique, where the gait classifier receives the top matches from the face classifier. Using data acquired from an optoelectronic motion capture system, a new system is developed for gait identification. This system makes use of a number of gait parameters that have been shown to be the most important for recognition. It is also feasible to identify someone without using their face, fingerprints, palm prints, knuckles, ears, or iris. Gait analysis allows for the recognition of a person even when the individual is not aware of it. Finding a person's gait characteristics is the goal of this study, which employs a CNN model, a deep learning algorithm. Python was the language of choice for the investigator as they built the foundational architecture of the deep learning CNN. The investigator's job is made easier by importing the default library packages in keras, an open-source neural network library designed in Python that operates on Tensorflow. The database is fine-tuned using industry-standard techniques such as grouping technic, systematic random picture removal, Laplacian fuzzy image removal, and contour broken image removal. Substituting a mix of the softmax and sparsemax functions for the softmax function in the last layer of the deep learning CNN considerably improves it.