Keras Based Real-time Hand Gesture Recognition

Abstract

Everyone can communicate effectively with hand gestures. Hand gesture recognition, as opposed to the traditional text-based or GUI-based method, can be viewed as a way for computers to better grasp the language of the human body and develop richer human-computer connections. Hand gestures can help to simplify tasks by allowing you to give orders to the computer using simple hand movements. The camera records gestures, the computer interprets them, and an appropriate action is executed in response to the gesture. This project is based on hand gesture recognition, which captures hand motions and converts them into instructions that the machine can comprehend and execute. In this project, we will first capture the gesture with the camera. The captured image is then preprocessed for feature extraction, and a convolutional neural network is implemented using a backpropagation algorithm to effectively recognize gestures and train the machine.