Object Detection Using Deep Learning (DL) and OpenCV Approach

Abstract

The main aim of the current research is to create a new kind of human-computer interaction system (HCIS) that addresses the issues that users have been facing with the existing object detection systems. We have proposed a method for real-time hand gesture detection using TensorFlow and Python programming employing OpenCV. In this approach, the image is taken and sent as a test image to the trained model dataset, where the pre-trained model will check the test image and try to classify based on the image regarding what kind of gesture it is and display the gesture name with the accuracy found in the image. In this context, hand gestures are the default mode of interaction when two people are speaking to one another, where hand movements can be considered a verbal method of communication. The training dataset of images that is used has five gestures, each with 20–30 variations of a single gesture with different angles, different people, and different backgrounds. The aim of doing this is to improve the accuracy of the dataset classification. Hand gestures are important mortal-to-mortal communication channels that convey a major part of information transfer in our everyday lives. The algorithm employed can reprocess numerous hand varieties while detecting the hand motions according to the test and train images and is resistant to changes in backdrop image because it isn't based on background image deduction and wasn't developed for a particular hand type. The application is able to do gesture recognition in real time.