Welcome to VisionFlow, a powerful computer vision system designed for internal use at a leading (but anonymous 👀) beverage company. VisionFlow ensures consistent quality, detects visual anomalies in production, and assists with inventory monitoring — all with the magic of machine learning and real-time video processing.
⚠️ This repository is for demonstration purposes. Brand-specific content and assets have been anonymized.
- Real-Time Object Detection
- Detect bottles, cans, labels, and caps on the production line
- Anomaly Detection
- Flag missing labels, cap misalignment, and fill level inconsistencies
- Inventory Insight
- Count units, track SKUs, and generate visual logs for auditing
- Performance Dashboard
- Live metrics with alerts and historical tracking (optional web UI)
| Layer | Technology |
|---|---|
| Language | Python |
| Model Framework | PyTorch / TensorFlow |
| Detection Model | YOLOv8 |
| Backend | FastAPI |
| Stream Handling | OpenCV |
| Dashboard (optional) | Streamlit or React |
git clone https://github.com/your-username/visionflow.git
cd visionflow
pip install -r requirements.txt- Clone the repository:
git clone https://github.com/yourusername/azvision.git
cd azvision- Create a virtual environment (recommended):
python -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate- Install dependencies:
pip install -r requirements.txt- Run the application:
python -m azvision-
Select your camera and resolution from the settings panel.
-
Adjust edge detection or color detection settings as needed.
-
Use the camera calibration tool to set up accurate measurements.
-
Capture images and generate DXF files for CNC machining.
- The application supports both DirectShow and FFmpeg cameras
- Common resolutions (640x480, 1280x720, 1920x1080, 2560x1440) are supported
- Camera settings can be adjusted in real-time
- Contours are automatically detected and simplified
- The DXF file is generated with proper scaling based on calibration
- Output is in inches with decimal precision
- Contours are closed polylines suitable for CNC machining
The application is structured into several modules:
azvision/gui/: GUI components and main applicationazvision/utils/: Utility functions for image processing and camera handlingazvision/calibration/: Camera calibration functionalityazvision/config.py: Configuration settings
- Fork the repository
- Create a feature branch
- Commit your changes
- Push to the branch
- Create a Pull Request
This project is licensed under the MIT License - see the LICENSE file for details.