Train with another dataset

[alikaz3mi]

Hello, I know that you have said to ask questions about the dataset from https://github.com/clovaai/deep-text-recognition-benchmark repository. I have checked their repo. It is mentioned that create your own dataset, "put images in one folder, create their gt.txt". in gt.txt there is no sign of bounding boxes. I'd like to fine-tune both text detection and text recognizer. How I can do that?

Thank you in advance.

[kimlia545]

@alikaz3mi
"https://github.com/clovaai/deep-text-recognition-benchmark" repository is the place for the character recognition model.
Here for text detection.
"https://github.com/clovaai/CRAFT-pytorch"
But the code for training is not included in this repository, and they cannot release the full training code for IP reason.

[saribolagh]

I want to fine-tune the https://github.com/JaidedAI/EasyOCR/releases/download/v1.3/english_g2.zip model but there are specific words in my data that I dont think there are in the this model. Which file should I modify? Also, do I need to convert the images and labels into .mdb, like in the repo https://github.com/clovaai/deep-text-recognition-benchmark or it is not necessary ?

[alikaz3mi]

@kimlia545
Thank you for your response.

[Manishsinghrajput98]

complete custom easy ocr recognition model training tutorial

https://www.youtube.com/watch?v=-j3TbyceShY