Hand Gesture Recognition and Control for Human-Robot Interaction Using Deep Learning

Abstract

This paper introduces a real-time system for recognizing hand gestures using Python and OpenCV, centred on a Convolutional Neural Network (CNN) model. The primary objective of this study is to address the challenge of recognizing hand gestures in varied and complex environments. The proposed approach employs several image and video processing techniques, including data augmentation and feature extraction, to segment the hand region and extract relevant features. The system’s performance is significantly improved by adding to the original training dataset, resulting in 5,000 images with 500 images per gesture, as shown by the evaluation metrics indicating a substantial increase in accuracy from 96.9% to 99.2%. This paper aims to provide feasible and economical solutions for utilizing robots in industrial settings, while also proposing future research possibilities for enhancing human-robot interaction through methods such as incorporating hand gesture recognition.