ALPR: Automatic License Plate Recognition System
ALPR (Automatic License Plate Recognition) is a comprehensive real-time system for detecting vehicle license plates in images or video streams and extracting the plate text via OCR. The project includes dual user interfaces tailored to two distinct roles.
Pipeline
- Detection: a fine-tuned object detection model (YOLO-family or Faster R-CNN) localises license plate bounding boxes in the input frame.
- Crop & Preprocess: detected regions are cropped, perspective-corrected, and contrast-enhanced for better OCR accuracy.
- OCR: an optical character recognition engine (Tesseract / custom CNN) reads the plate characters.
- Database lookup: recognised plates are matched against a PostgreSQL database of registered vehicles.
- Action: flag unknown plates, log recognised entries, and trigger alerts.
Dual GUI
- Security Manager interface: real-time video feed with overlaid detections, plate text readout, match history, and alert management.
- Car User interface: a simpler view showing recognition status, entry logs, and personal vehicle records.
Both interfaces built with Streamlit for rapid deployment without a custom frontend.
Technology
Python, PyTorch, OpenCV, Tesseract OCR, PostgreSQL, Streamlit.