A high-performance Human-Computer Interaction (HCI) system that uses real-time hand-landmark detection to control your operating system. Built with modern computer vision standards for low-latency cursor control.
- Modern Tasks API: Leverages the 2024 Mediapipe Tasks Vision for robust, real-time hand tracking.
- Hardware Acceleration: Specifically optimized for Apple Silicon (M3) using the
tensorflow-metalframework. - Pinch-Precision Interaction: Smart gesture recognition for "Pinch-to-Click" with distance-based normalization.
- Motion Smoothing: Implements a jitter-reduction algorithm for ultra-smooth cursor navigation.
- Computer Vision: Mediapipe
HandLandmarkerTasks API. - System Automation: PyAutoGUI for cross-platform OS-level mouse and keyboard control.
- Signal Processing: Moving average filters and Euclidean distance-based gesture logic.
- Low Latency: Optimized for real-time video capture at 30+ FPS on ARM-based hardware.
- Local Setup:
git clone https://github.com/AkkiKrsingh2005/ai-gesture-os-control.git cd ai-gesture-os-control pip install -r requirements.txt python main.py - Gesture Guide:
- Index Movement: Direct cursor mapping.
- Thumb + Index Pinch: Native Left Click.
- Press 'q': Exit.
Developed by Ankit Kumar | Portfolio