A Vision-based Gesture Identification Framework for Real-Time Device and System Control

In this paper a real time gesture-controlled device automation system has been provided which is an integration of computer vision and embedded hardware that has been used to have efficient and contactless control. The suggested system uses hand landmark detection with MediaPipe and OpenCV to take up images and Arduino microcontroller to integrate hardware reaction, respectively. The system identifies and processes the hand gestures, which it captures with a web camera and maps them to the set control commands of the software and the peripherals. Python and a 30 FPS webcam were used to implement the experiment with different light levels under conditions of experimental responsiveness and recognition accuracy. The system reported an average gesture recognition rate of 96.8 percent and a response time of less than 0.1 seconds which indicates that the system is viable in real time. These findings confirm the effectiveness of MediaPipe in performing a gesture recognition with high accuracy and the stability of Arduino-based hardware implementation of a multimodal control. The paper emphasizes the scalability of the system to be used in automation applications, accessibility technologies and smart environments.