ninv / korean-license-plate-recognition Goto Github PK

Thank you for your project.

When I test 1 line plate image in your test_images, the result is quite good.

But when I test 2 line plate image in your test_images, the result is not correct.

Someone says that LPRNet is for 1 line, so we cannot train LPRNet for 2 line plate.

As you know, there are many kinds of License plate in Korea.

Is it possible to train LPRNet for 2 line plate image?

I’m trying to train your code on my own dataset. But your labeling style is kind of different. What labeling tool did you use? Or I’m familiar with tzutalin/labelImg. How can I change your code to be trained on pascalvoc format dataset?

hi there.
I tried to download best_weights.zip, but cannot download.
can you send file to my e-mail, or let me know another download link?
thanks. [email protected]

I'm trying to train the model with Variable length labels. While the model trains well with batch size 1, I run into an error when using batch size >= 2

Cannot batch tensors with different shapes in component 1. First element had shape [9] and element 1 had shape [10]. [Op:IteratorGetNextSync]

Thank you for the code.

I successfully run your all codes. But when I train it with your given data, it returns empty vector. What could be the possible problem?

Is it possible to share KarPlate dataset? If possible, please share about the number of training/testing data.

your model's network architectureis are different from paper's

The training database can not be downloaded due to legal problems from here.
http://pr.gachon.ac.kr/ALPR.html
So I don't know what should be the format of the json label file for the training.

Hello there can you please help me to overcome this error

I'm just trying to run the demo

Hi, how can I get a graph of accuracy?