How nadir199/pygesture-side-project-pytorch-scholarship-challenge works
Main Components
a) Frontend Layer:
• Sub-components for:
• Templates (home.html, settings.html, layout.html)
• Static assets (CSS, JS, bootstrap files)
b) Backend Layer:
• Flask Server component (index.py)
• Gesture Recognition Engine (program.py)
• Neural Network Model (net.pt)
• Settings Manager (settings.json)
c) Hardware Integration:
• Webcam Input component
• System Output (keyboard/browser actions)
Component Relationships
• User → Web Interface → Flask Server → Settings Manager (configuration flow)
• Webcam → Gesture Recognition Engine → Neural Network Model → System Actions (runtime flow)
Specific Implementation Details
A) Configuration Flow:
[Web Browser] → [Flask Server (index.py)]
• ↓
• [Settings Manager]
• (settings.json)
B) Runtime Flow:
[Webcam] → [Gesture Recognition (program.py)] → [PyTorch Model (net.pt)]
• ↓
• [Action Executor]
• ↓
• [System Actions (keyboard/browser)]
Important Notes
• A clear distinction between the configuration phase (web interface) and the runtime phase (gesture recognition)
• Two main gesture types (Fist, High five) and three action types (website, key press, phrase typing)
• Technology stack: Flask, PyTorch, OpenCV
• HTTP server port (5000) for the web interface
Additional Elements
• Data flow directions
• Component types
• Configuration vs Runtime modes
• Input/Output interfaces
Runtime workflow through the gesture recognition system
This will help other developers understand both how to configure the system and how it operates during runtime.