How do I add EasyOCR

[RageMist]

Hello Sir Xulihang...
Thanks for the Coupon I purchased the Software. I wanted to know how do I add and use EasyOCR in the ImagesTrans application, I couldn't find anything in the Documentation..

[xulihang]

Have you installed it?

1. You have to run it as a server using this python script: https://github.com/xulihang/ImageTrans_plugins/blob/master/easyOCR/server/server_easyocr.py
2. Put the plugin files for easyOCR in ImageTrans's plugins folder: https://github.com/xulihang/ImageTrans_plugins/

[RageMist]

I don't think its working properly, don't you have any guide for that......and I also saw that other than Azure and Tesseract no other OCR is working......

[RageMist]

Can you provide any solution for this......

[xulihang]

The OCR list is not complete. The OCR plugins are not loaded properly. How did you install ImageTrans? Are you using the package from OneDrive?

What image do you need to translate?

[RageMist]

There are more options but it have to be scrolled down...

[RageMist]

As for the image it a Webtoon.....I'll send you the the first page.....

ou the first page.....

[xulihang]

If you are using the package from onedrive, restart ImageTrans to see if you can see OCR plugins like WinRT, Google, paddleocr, mangaocr, etc.

[RageMist]

Ocrspace doesn't do proper OCR, Tesseract and Azure are 50-50 sometimes they don't detect text sometimes only half of the text, no responce from youdao, abby, clova, google,tencent, winRT,paddleOCR and all baidu response to "null".....and MangaOCR only does japanese and I need Chinese.......

ImageTrans__2023-09-24_20-11-18.mp4

[RageMist]

So! Sir Xulihang, What should I do now.....

[xulihang]

Not all OCRs are available by default as I wrote in the docs: https://imagetrans.readthedocs.io/en/latest/gettingstarted.html#ocr

You have to apply for the API keys to use them, or install and run the engines locally.

Baidu is a service by a Chinese company. It recognizes Chinese accurately. But it seems that it is blocked in India. Youdao and tencent are also Chinese services. They are not applicable in India.

You can try to use Google.

There are two ways to use it.

1. A free option is to use Google Drive. You can use it in ImageTrans directly or run it by yourself: Multiple issues #432 (comment)
2. Apply for the API key for Google Vision OCR according to the link in the FAQ page: https://www.basiccat.org/imagetrans/faq/

Of course, there are other options. But they may not be as accurate as Google.

[xulihang]

Here is a video using Google Vision OCR:

google_ocr.mp4

[xulihang]

WinRT is Windows 10's built-in OCR. It works great for Chinese as well and it is free as it runs on your device.

It seems that it is not called correctly. Maybe the language models are lacking. Try to install Chinese as the system's language.

Then you can use it as in the video:

winrt.mp4

[RageMist]

I'll give Google a shot, but just in case it don't work for me is there any option to refund or something.....

[xulihang]

No problem.

[RageMist]

I have another Question if you are still avalible, I want use ChapGPT as Machine Translation because its far betten than the rest but even after add the API key it still doesn't response........

[xulihang]

Check out this blog about ChatGPT: https://www.basiccat.org/ChatGPT-image-translator/

Use ImageTrans.bat to start ImageTrans and you can check out its outputs to see why it fails.

[RageMist]

I did follow the blog but it still doesn't work for me....I looked at the Google Vision, and its free plan only allows 1000units/month, which might be around 20-25 manhua chapters......The ImagesTrans Application is not bad but maybe for me, it's a bit complex and not suitable.....I hope you can issue a refund. And I'm Sorry for all the trouble....

[RageMist]

Sir Xulihang, Do I need to send a Mail with a Refund Request?

[xulihang]

Refunded. Please check your account.

[RageMist]

I live in India, So I'm sure it will take a few days to credit. but Thank You!