Motivation

ALPR and ANPR

Automatic License Plate Recognition

Automatic Number Plate Recognition

I wanted to build out a mobile ALPR app without paying a subscription cloud fee to someone. I dug in and found that it is a pretty layered problem, especially with the constraints I had given myself.

Datasets

A few different ones

https://www.kaggle.com/datasets/tolgadincer/us-license-plates

https://www.plateshack.com/y2k/index.html

https://public.roboflow.com/object-detection/license-plates-us-eu?utm_source=chatgpt.com

https://www.kaggle.com/datasets/gpiosenka/us-license-plates-image-classification/data

Tesseract v PaddleOCR

Two different big names in opensource OCR.

Tessdata

309 KB, from Dec 2015. https://github.com/SandroMachado/openalpr-android/blob/master/app/src/main/assets/runtime_data/ocr/tessdata/lus.traineddata

https://tesseract-ocr.github.io/tessdoc/Data-Files

https://github.com/tesseract-ocr/tessdata/raw/4.00/eng.traineddata

Other runtime data

Keypoints, Postprocess, Region, etc.

https://github.com/SandroMachado/openalpr-android/tree/master/app/src/main/assets/runtime_data

Raw Image to License Plate number

https://www.ultralytics.com/blog/using-ultralytics-yolo11-for-automatic-number-plate-recognition

YOLOvXXX, OpenCV, Darknet

https://github.com/Hyuto/yolov8-tfjs

https://github.com/AlexeyAB/darknet

https://github.com/christianroman/ANPR

https://github.com/aqntks/Easy-Yolo-OCR

https://medium.com/saarthi-ai/how-to-build-your-own-ocr-a5bb91b622ba

https://medium.com/@bibekdhakal.dev/an-introduction-to-license-plate-recognition-with-python-a9a69ac4a06d

https://github.com/Dodant/anpr-with-yolo-v4?tab=readme-ov-file

https://github.com/theAIGuysCode/yolov4-custom-functions/blob/master/data/video/cars.mp4

https://www.youtube.com/watch?v=AAPZLK41rek&t=3s

https://www.reddit.com/r/computervision/comments/ixmncf/license_plate_recognition_using_yolov4_opencv_and/

Yolo and Darknet need python. So this is running on a server with some good graphics cards behind it.

This would be difficult to get into a mobile device today...

Stackoverflow

https://stackoverflow.com/questions/19268648/using-tesseract-to-recognize-license-plates

Tensorflow, Nvidia

https://www.kaggle.com/code/thnhhiu/plate-detector-using-tensorflow/notebook

https://developer.nvidia.com/blog/creating-a-real-time-license-plate-detection-and-recognition-app/

Fonts

https://www.1001fonts.com/license-plate-fonts.html

https://ia800802.us.archive.org/1/items/license-plate-fonts/USALicensePlate.ttf

openalpr

https://github.com/openalpr/train-detector

https://github.com/openalpr/train-ocr/blob/master/train.py

npm options, tesseract.js tessseract-wasm yolov8-tfjs

https://transloadit.com/devtips/integrating-ocr-in-the-browser-with-tesseract-js/

https://github.com/robertknight/tesseract-wasm

https://github.com/naptha/tesseract.js/