University-Vehicle-Tracker

An integrated system for vehicle tracking and identification on university campuses. This system combines license plate recognition and facial recognition to identify vehicles and their occupants in real-time for Automated Attendance Tracking.

Features

• License Plate Detection: Detects both rectangular and square license plates
• License Plate Recognition (LPR): Extracts and recognizes characters from license plates
• Face Detection: Locates faces in video frames
• Face Recognition: Identifies known individuals from detected faces
• Real-time Processing: Processes video stream with FPS counter

Project Structure

University-Vehicle-Tracker/
├── Character-Time-series-Matching/  # License plate recognition components
│   ├── Char_detection_yolo.py       # Character detection using YOLOv5
│   ├── process_plate.py             # License plate processing
│   ├── evaluate.py                  # Evaluation script
│   └── character_name.txt           # Character classes
├── face-recognition/                # Face recognition components
│   ├── detect.py                    # Face detection module
│   ├── recognize.py                 # Face recognition module
│   ├── main.py                      # Main face recognition application
│   └── insightface/                 # Face embedding model
├── detection.py                     # Main application script
├── requirements.txt                 # Project dependencies
└── README.md                        # Project documentation

Installation

1. Clone the repository:

git clone https://github.com/yourusername/University-Vehicle-Tracker.git
cd University-Vehicle-Tracker

2. Install dependencies:

pip install -r requirements.txt

3. Download required model weights:
   • Character detection models: object.pt and char.pt (place in Character-Time-series-Matching/Vietnamese/)
   • Face detection model: yolov5n-0.5.pt (place in face-recognition/yolov5_face/)
   • Face recognition model: resnet100_backbone.pth (place in face-recognition/insightface/)

Usage

Run the main detection script:

python detection.py

The system will:

1. Access your webcam
2. Detect and recognize license plates
3. Detect and recognize faces
4. Display results in real-time with bounding boxes and labels

To exit the application, press 'q'.

Dependencies

• PyTorch
• OpenCV
• NumPy
• SQLAlchemy
• TorchVision
• YOLOv5 (included)
• InsightFace (included)

Model Information

• License Plate Detection: YOLOv5 model trained on license plate data
• Character Recognition: YOLOv5 model trained on license plate characters
• Face Detection: YOLOv5-face model (yolov5n-0.5.pt)
• Face Recognition: iresnet100 backbone for face embedding

License

Refer to the LICENSE file for licensing information.

TODOS

1. Increase FPS
2. Improve character mapping on indian license plates
3. Store recognized input in sqlite database
4. Add Jetson Nano support