search

stephanecharette / DarkPlate

License plate parsing using Darknet and YOLO
MIT License
58 stars 11 forks source link

More details #1

Open folkien opened 3 years ago

folkien commented 3 years ago

Could you explain better the idea behind this projected? As far I am concerned, it looks like you :

  1. Trained YOLO to recognize (Letters/Numbers + License Plate)
  2. Parse video, detect each frame
  3. Filter only detections inside 'License Plate detection bbox'
  4. Read detections labels from left to right
stephanecharette commented 3 years ago

Correct. The process I was showing with this project is to use the same YOLO network twice. Once to find the plate, then again to read the characters. No need to have 2 networks, or use tesseract to "read" the plates.

stephanecharette commented 3 years ago

This was part of a long discussion that took place on the Darknet/YOLO discord, come join if you want more details: https://discord.gg/zSq8rtW