ANPR-ATCC is a powerful, unified traffic monitoring platform combining:
- Automatic Number Plate Recognition (ANPR): Real-time detection and OCR extraction of vehicle license plates from images, videos, and camera streams.
- Automatic Traffic Classification (ATCC): Multi-class vehicle detection supporting traffic density classification from live or recorded media.
This solution leverages the best of modern computer vision and OCR technologies, providing a scalable, flexible, and easy-to-use system ready for deployment in smart city projects, traffic analytics, law enforcement, and academic research.
Accurate YOLOv10 detection of diverse license plates combined with Tesseract OCR for high-fidelity text extraction.
This feature enables real-time, prioritized security and access control based on the detected license plate against two managed lists:
The system instantly recognizes plates belonging to:
- VIP personnel (e.g., government officials)
- Emergency services (ambulances, police, fire trucks)
Detection of a whitelisted plate triggers:
- Immediate, high-priority alert
- Optional integration with access control systems (e.g., boom barriers)
- Swift automated clearance through checkpoints or controlled zones
A continuously updated list of stolen, flagged, or wanted vehicles.
When a blacklisted plate is detected:
- A critical violation alarm is triggered
- Location and timestamp are logged
- Optional real-time notification to law enforcement for interception
YOLOv11n-based vehicle detection that classifies vehicle types, counts them, and estimates traffic congestion in real time.
Supports:
- Video files
- Images
- Real-time camera streams
- Experimental: Webcam streaming
Outputs are persistently stored in:
- SQLite databases
- Cumulative JSON files
Enhanced Streamlit dashboards provide:
- ATCC traffic visualizations
- Traffic flow analysis
- Peak-hour heatmaps
- Trend analysis
- Historical detections overview
Streamlit-powered interface with:
- Confidence and IoU threshold sliders
- Real-time detection previews
- Built-in database management tools
Includes:
- Graceful fallback if Tesseract OCR is unavailable
- Temporary file cleanup
- Reliable database synchronization
Tested on:
- Linux
- Windows
- macOS
Comes with detailed environment setup instructions.
Separate pipelines for:
- ANPR
- ATCC
This ensures:
- Easy maintenance
- Clean codebase
- High extensibility
| Technology | Description | Justification |
|---|---|---|
| YOLOv10 & YOLOv11n (Ultralytics) | Ultra-fast, state-of-the-art object detection architectures. | Proven accuracy and speed, customizable weights for license plates and vehicle types. |
| Tesseract OCR | Open-source text recognition engine supporting multiple languages. | Lightweight, widely supported, best integration with Python workflows. |
| Streamlit | Python framework for building interactive data apps. | Rapid prototyping with minimal code for highly interactive UIs. |
| SQLite | Lightweight, serverless SQL database. | Perfect for embedded analytics, portable, zero-config, easy integration with Pandas. |
| OpenCV | Image and video processing library. | Industry-standard computer vision operations with Python bindings. |
| pandas & matplotlib | Data manipulation and visualization libraries. | Powerful data analytics and comprehensive plotting capabilities. |
| Python Standard Library | Utilities for file I/O, regex, system processes, date/time handling. | Robust tooling for supporting application logic. |
- Municipal Transportation Departments: Traffic flow monitoring and violation detection.
- Parking Facility Operators: Automated license plate-based entry/exit logging.
- Law Enforcement: Quick violation checks and real-time surveillance analytics.
- Researchers & Academia: Traffic pattern analysis & machine learning datasets collection.
- Smart City Solutions Providers: Traffic management dashboards integrated with IoT devices.
- Educators & Students: Open-source educational project for computer vision and data science.
- Python 3.11+
- Tesseract OCR installed and added to system PATH
- Conda environment recommended for dependency isolation
git clone https://github.com/KomatiBhavaniSankar/ANPR-ATCC-Infosys.git
cd ANPR-ATCC-Infosys
conda create -n cvproj python=3.11 -y
conda activate cvproj
pip install --upgrade pip
pip install -r requirements.txt
cd yolov10
pip install -e .
cd ..
# Ensure SQLite databases are created automatically, else run:
python sqldb.pyIf CUDA-enabled GPU is available, install torch and ultralytics versions compatible with your CUDA version to accelerate YOLO models.
TESSDATA_PREFIX(optional): Path to Tesseract language data files if custom installed.
- Fix numpy version conflicts:
pip uninstall numpy pip install numpy==1.26.4
- Confirm tesseract CLI works via terminal by running:
tesseract --version - Check camera permissions on your OS before using webcam inputs.
- Upload video/image or use webcam to detect vehicle plates in real-time.
- OCR extracted plate text displayed, saved to SQLite DB and JSON snapshots every 20 seconds.
- Visual overlays help easily identify bounding boxes and recognized characters.
- Upload image/video or camera capture to detect various vehicle classes (cars, trucks, bikes).
- View detailed analytics including vehicle counts, traffic levels (No/Low/Medium/High), and historic data logs.
- Adjust confidence and IOU thresholds dynamically to tune detection sensitivity.
- View raw analysis tables with full timestamped historic records in-app.
- Clear or reset databases with UI buttons.
- Export DB files for offline analysis or import into third-party tools.
- Fork and clone the repo.
- Create a new feature or bugfix branch.
- Follow Python style conventions (PEP8) and write descriptive commit messages.
- Test your changes thoroughly with provided notebooks and media.
- Document any new APIs, workflows, or configurations in README or docstrings.
- Submit a pull request referencing related issues or features.
- Example media files and Jupyter notebooks are provided under
notebook/for functional validation. - Unit tests for database operations and core detection pipelines to be added.
- Continuous integration support planned (GitHub Actions) for automatic test runs on pull requests.
- All data stored locally—no external servers involved, keeping sensitive information secure by design.
- For production deployments, consider encrypting SQLite DB files and securing access paths.
- Webcam and file uploads handled within browser sandbox, with no persistent external uploads.
- Future versions may integrate authentication layers to enable controlled multi-user access.
- Multi-language OCR support: Increase scope beyond English license plates with PaddleOCR and custom models.
- Integration with Vehicle Registration APIs: Cross-check extracted plates with official databases for real-time alerts.
- Edge Device Deployment: Lightweight models optimized for Jetson Nano, Raspberry Pi, or embedded cameras.
- Cloud Sync & Visualization: Remote data dashboard on cloud platforms (AWS/GCP/Azure).
- Enhanced UI/UX: Dark mode, user profiles, notifications, and mobile responsiveness.
- Expanded Vehicle Classification: Include electric scooters, buses, emergency vehicles, and non-motorized entities.
- AI Model Improvements: Experiment with Transformer-based detectors for superior accuracy.
- YOLOv10 GitHub
- Ultralytics YOLO Documentation
- Tesseract OCR
- Streamlit Official Site
- SQLite Browser
- Python OpenCV
This project is licensed under the MIT License © 2025 Vidzai Digital.
ANPR-ATCC aims to revolutionize intelligent traffic monitoring by merging AI-powered vision, automation, and real-time analytics into one unified platform — contributing toward smarter, safer, and more efficient cities.