Gaming with bare hands
- a machine learning approach
This was my Final Year group Project in University. The objective of the project is to develop a software that can act as a gaming controller that utilizes hand gesture motion control. This project will investigate existing machine learning based hand gesture recognition algorithms and improve it to effectively play a computer games such as racing car, card games, or the alike. I worked on the Deep learning model of the Project, utilising MediaPpipe's hand landmark detection framework, TensorFlow and OpenCV. To process the hand gesture images that I input and extract the hand landmarks, and train a simple feedforward neural network for classification. Then the pre-trained model is loaded to a controller that recognizes hand gestures via webcam and maps them to the game controls using pydirectinput library.
