ALPR Indonesia

Automatic license plate recognition for Indonesian plate (White on black)

This code was the adjustment version from :
https://github.com/MicrocontrollersAndMore/OpenCV_3_License_Plate_Recognition_Python

Methods:

Object detection is manualy segmenting plate, and recognize each candidate number or alphabet using knn method.
Here for the detail :

• Video Explanation
• KNN
  

Note :
It is recommended to use a newer method like yolo or ssd.
or you can read the state of the art of object detection
State-of-the-art of Object Detection

Prerequest

• python==3.6
• opencv-python==4.1.1.26
• numpy==1.17.3
  or install using
  pip install -r requirements.txt

How to run (Linux):

• Calibrate the camera and threshold
  python Main.py -c <image_dir>
• Cam
  python Main.py
• Image
  python Main.py -i <image_file_dir>
• Video
  python Main.py -v <video_file_dir>

Retrain

Retrain process will update classifications.txt and flattened_images.txt files
python GenData.py -d = <train_image>
example :
python GenData.py -d = train_image/train2.png
note: Just input base on marked object one by one and press esc to exit the training process

Check the model

python TrainAndTestData.py -d = train_image/train2.png

Tools

• Invert image:
  python invert_imageData.py -d = train_image/train2.png