License_plate_recognition_example

kyushu / Computer_Vision

Computer Vision

0 stars 0 forks source link

License_plate_recognition_example #1

Open rutujar opened 5 years ago

rutujar commented 5 years ago

Nice work on License plate. can i know in train_advanced.py the argument to run the program is python train_advanced.py --samples output/examples --char-classifier output/adv_char.cpickle \

--digit-classifier output/adv_digit.cpickle but in the code there are 4 argument parser called

samples, char-classifier , digit-classifier, min_samples. why? can i know what is the limitations in this project.

kyushu commented 5 years ago

@rutujar if you want to do something about LPR, i think you should go Deel Learning like CNN, RNN .etc.. here is a reference https://github.com/qjadud1994/CRNN-Keras

This project uses computer vision/Image processing to locate License plate, segment char and extract feature of image patch of char for training a SVM classifier model.

Here is summary of this project

The accuracy of locate License Plate and segment char are highly affected by light condition.
The accuracy is not good enough. it is about 7x % ~ 80 %.
Hard/annoying to tune hyper parameters.

rutujar commented 5 years ago

@kyushu Thanks, i am working on SVM itself.

rutujar commented 5 years ago

will this code for 2 license plate in an image,or is there any restriction so that i can modify it..

kyushu commented 5 years ago

you may check mtImageSearch/license_plate/license_plate.py. LicensePlateDetector detects all possible license plate region in the image.

rutujar commented 5 years ago

@kyushu this code works in python2 or python3?