Add a new OCR model

[Ivachnenco]

Google Lens stopped working again, which is a shame since it was the best model for capturing English text in manga in the program. Therefore, is it possible to add more sources of text recognition? The same PaddleOCR for example, which is used in luna translator and recognizes English letters perfectly. And also add Ocrspace which can be used thanks to a free API key. It also works very well and can be an alternative to google lens as an online service

[elfcute]

Can large model OCRs like Gemini 2.0 flash-exp, DeepSeek be added?

[abysmli]

could you please add google vision (paid version) plugin?

[bropines]

Can large model OCRs like Gemini 2.0 flash-exp, DeepSeek be added?

I have a module ready, but I'm too lazy to release it.

[elfcute]

Can large model OCRs like Gemini 2.0 flash-exp, DeepSeek be added?

I have a module ready, but I'm too lazy to release it.

You're too good

[elfcute]

Can large model OCRs like Gemini 2.0 flash-exp, DeepSeek be added?

I have a module ready, but I'm too lazy to release it.

Hi, I have another question. Do you remember that I contacted you on Telegram previously? One of the issues I brought up was that some large language models, like Gemini (gemini-2.0-flash-exp, gemini-exp-1206, gemini-2.0-flash-thinking-exp), sometimes encounter errors when translating text from a single image dialog box, especially when there is a large amount of text within that single dialog box.

Is this something that can be optimized and resolved?

Thanks.

[bropines]

Can large model OCRs like Gemini 2.0 flash-exp, DeepSeek be added?

I have a module ready, but I'm too lazy to release it.

Hi, I have another question. Do you remember that I contacted you on Telegram previously? One of the issues I brought up was that some large language models, like Gemini (gemini-2.0-flash-exp, gemini-exp-1206, gemini-2.0-flash-thinking-exp), sometimes encounter errors when translating text from a single image dialog box, especially when there is a large amount of text within that single dialog box.

Is this something that can be optimized and resolved?

Thanks.

From a part. I plan to use structured responses, but unfortunately not all apis are able to work with them.
The previous logic worked based on the fact that we stupidly order the neural network to follow the rules, which it constantly violated. Google, openai, and claude have structured response functions, but the problem is that this needs to be integrated into the current translation method + a new ocr engine needs to be added. I'm not really lazy. It's just that I'm a student, and I don't have time to finish the modules during the session. After I get some rest, I'll make a new ocr module and update the translator.

[elfcute]

Can large model OCRs like Gemini 2.0 flash-exp, DeepSeek be added?

I have a module ready, but I'm too lazy to release it.

Hi, I have another question. Do you remember that I contacted you on Telegram previously? One of the issues I brought up was that some large language models, like Gemini (gemini-2.0-flash-exp, gemini-exp-1206, gemini-2.0-flash-thinking-exp), sometimes encounter errors when translating text from a single image dialog box, especially when there is a large amount of text within that single dialog box.
Is this something that can be optimized and resolved?
Thanks.

From a part. I plan to use structured responses, but unfortunately not all apis are able to work with them. The previous logic worked based on the fact that we stupidly order the neural network to follow the rules, which it constantly violated. Google, openai, and claude have structured response functions, but the problem is that this needs to be integrated into the current translation method + a new ocr engine needs to be added. I'm not really lazy. It's just that I'm a student, and I don't have time to finish the modules during the session. After I get some rest, I'll make a new ocr module and update the translator.

Oh, you're a student. You're amazing. You've been working so hard recently. Once you're done with your studies, make sure to rest well and recharge. We are all really looking forward to your new OCR engine.

[bropines]

By the way. I already added an ocd module a couple of days ago, based on Gemini and OpenAI. Test it out

[Bubucenter]

By the way. I already added an ocd module a couple of days ago, based on Gemini and OpenAI. Test it out

Hi thanks for adding ocr openai. I hope you can add 3rd party api url. So i can add 3rd party API. Thanks you very much

[bropines]

By the way. I already added an ocd module a couple of days ago, based on Gemini and OpenAI. Test it out

Hi thanks for adding ocr openai. I hope you can add 3rd party api url. So i can add 3rd party API. Thanks you very much

Ummm. He's there. if your third-party API supports OpenAI SCHEMES, then everything will work, the main thing is to insert the link to the endpoint.

[Bubucenter]

By the way. I already added an ocd module a couple of days ago, based on Gemini and OpenAI. Test it out

Hi thanks for adding ocr openai. I hope you can add 3rd party api url. So i can add 3rd party API. Thanks you very much

Ummm. He's there. if your third-party API supports OpenAI SCHEMES, then everything will work, the main thing is to insert the link to the endpoint.

I added your OpenAI translator. It works great. Hopefully the OCR part will do the same.

[bropines]

By the way. I already added an ocd module a couple of days ago, based on Gemini and OpenAI. Test it out

Hi thanks for adding ocr openai. I hope you can add 3rd party api url. So i can add 3rd party API. Thanks you very much

Ummm. He's there. if your third-party API supports OpenAI SCHEMES, then everything will work, the main thing is to insert the link to the endpoint.

I added your OpenAI translator. It works great. Hopefully the OCR part will do the same.

It's basically no different from the translator, except that it sends pictures and promt. When you test it, write back, we'll close the issue

[Bubucenter]

By the way. I already added an ocd module a couple of days ago, based on Gemini and OpenAI. Test it out

Hi thanks for adding ocr openai. I hope you can add 3rd party api url. So i can add 3rd party API. Thanks you very much

Ummm. He's there. if your third-party API supports OpenAI SCHEMES, then everything will work, the main thing is to insert the link to the endpoint.

I added your OpenAI translator. It works great. Hopefully the OCR part will do the same.

It's basically no different from the translator, except that it sends pictures and promt. When you test it, write back, we'll close the issue
Can you create an ocr module that has components like this translate item?

[bropines]

By the way. I already added an ocd module a couple of days ago, based on Gemini and OpenAI. Test it out

Hi thanks for adding ocr openai. I hope you can add 3rd party api url. So i can add 3rd party API. Thanks you very much

Ummm. He's there. if your third-party API supports OpenAI SCHEMES, then everything will work, the main thing is to insert the link to the endpoint.

I added your OpenAI translator. It works great. Hopefully the OCR part will do the same.

It's basically no different from the translator, except that it sends pictures and promt. When you test it, write back, we'll close the issue
Can you create an ocr module that has components like this translate item?

Apparently you didn't really want to figure it out.

[Bubucenter]

Nhân tiện. Tôi đã thêm một mô-đun ocd vài ngày trước, dựa trên Gemini và OpenAI. Hãy thử nghiệm nhé

Xin chào, cảm ơn vì đã thêm ocr openai. Tôi hy vọng bạn có thể thêm url api của bên thứ 3. Vì vậy, tôi có thể thêm API của bên thứ 3. Cảm ơn bạn rất nhiều

Ừm. Anh ấy ở đó. Nếu API của bên thứ ba của bạn hỗ trợ OpenAI SCHEMES thì mọi thứ sẽ hoạt động, điều quan trọng nhất là chèn liên kết đến điểm cuối.

Tôi đã thêm trình dịch OpenAI của bạn. Nó hoạt động rất tốt. Hy vọng phần OCR cũng sẽ hoạt động như vậy.

Về cơ bản thì nó không khác gì trình dịch, ngoại trừ việc nó gửi hình ảnh và promt. Khi bạn kiểm tra, hãy viết lại, chúng tôi sẽ đóng vấn đề
Bạn có thể tạo một mô-đun ocr có các thành phần như mục dịch này không?

Có vẻ như bạn không thực sự muốn tìm hiểu vấn đề này.

I did it. Thank you for your guidance. Thank you very much.

[bropines]

I think it's time to close the issue.