Machine Learning Based Approach Using Hand Gestures for Mouse and Video Control

Machine-learning-based hand gesture recognition system will be greatly effective for individuals that help to increase utility without much human interference from input device at a distant or for the disabled. It will bring accuracy and hands-free control to experts and professionals, for example, in multimedia creation or virtual reality, hence improving workflow and interaction with digital environments. The suggested system is based on a real-time, low-cost hand gesture recognition system for HCI that uses CNN for hand detection, gesture segmentation, and classification. The existing systems employ models based on a framework that has used a convolutional neural network to create a video player controlled via human gestures. The existing system used a web camera to capture user gestures and performs basic activities like play and pause. Implemented in Python with Keras and Tensorflow, the system recognize and translate these gestures as video player commands. Deep learning techniques and Convolutional Pose Machines architecture are used for hand detection. The initialization procedures are carried out using Kalman filter and multi-frame recursion, hence reducing the number of redundant and erroneous frames. The proposed approach implements AI Virtual Mouse technology, which is based on capturing the hand movements, for controlling video and mouse operations using a web camera. Adopting OpenCV with Hand Detection and a Speech-To-Text library applied for implementing voice commands, the system is enabled so that users can navigate the virtual screen and manage video playback functions like play, pause, adjust volume, etc. The touch-free and multimodal approach improves. It has accessibility and provides a much more user-friendly experience.