Computer Vision LED Control System
This project demonstrates the integration of computer vision with embedded hardware by allowing LED flashes to be controlled using hand gestures. The system uses MediaPipe’s pre-trained hand detection model to recognize hand signals in real time, then sends the detected commands to an Arduino microcontroller responsible for controlling the LEDs. It was built as a practical exploration of how AI-powered perception can interact with physical electronic components.
The Challenge
Traditional hardware control usually depends on physical buttons, switches, or manual input. This limits interaction flexibility and does not provide a natural way for users to control devices without touching them. The project addresses this by exploring how computer vision can be used as an alternative input method for controlling hardware components.
The Solution
The solution uses a webcam-based computer vision pipeline to detect hand landmarks and gestures. Once a gesture is recognized, the software sends a command through serial communication to the Arduino. The Arduino receives the command and triggers the corresponding LED behavior, creating a simple but effective real-time interaction between AI perception and hardware control.
Architecture
The system follows a simple AI-to-hardware workflow. A camera captures live video frames, Python processes the frames using OpenCV and MediaPipe, hand landmarks are analyzed to detect user gestures, and the detected gesture is converted into a serial command. The Arduino receives this command and controls the LEDs based on the signal.