Ml Based Hand Gesture Recognition

Abstract

The projects focus on Gesture recognition using the media pipe hands Lite framework where the custom data set of 10 gesture Images is trained with an in built media pipe hands model which contains 2CNN models-A palm detection model running on single Shot detection architecture and a hand landmark generator running on regression model architecture. The data set is successfully trained and tested with the proposed method and an accuracy of 98 percent is obtained. Keywords: Gesture Recognition, Media pipe, Hands De-tection, CNN, Regression Model, Hand Landmark Detection, Machine Learning, Image Processing.