Skip to content

Latest commit

 

History

History
18 lines (14 loc) · 1004 Bytes

README.md

File metadata and controls

18 lines (14 loc) · 1004 Bytes
Raw

Romanian Plate Recognition System

See it in action: https://bit.ly/3BTuyx5

Plate recognition system designed to recognise Romanian number plates from photos and videos, using a custom in-house trained machine learning model, done in collaboration with Nokia's Innovation Labs.

Project featured on WUT's Faculty of Mathematics & Informatics website: https://www.info.uvt.ro/rezultate-studenti/

Plate Recognition from photos

  • Features:
    • Edge detection
    • Contour detection
    • Mask creation
    • easyOCR for character recognition

Plate Recognition recognition from videos

  • Features:
    • Custom in-house trained machine learning model to approximate the number plate's location. The model was trained with more than 1000+ images, assuring an accuracy of 96% rate of succes. The model returns the bounding box of the found license plate that's used for character recogntion
    • Pytesseract used for character recognition with character segmentation and morphological transformations.