Unrecognized areas

[bropines]

There are some pictures in which you don't want anything to be recognized. Like an author's signature, or a watermark. Is it possible to create a layer that will tell the OCR or Text detector not to detect here, move on?
Like the text at the bottom of this picture.

[lhj5426]

你可以先 OCR一下 然后把这个OCR结果 写入 OCR 替换 并启动 去掉无字框
这样就 没事 临时解决方案

[bropines]

你可以先 OCR一下 然后把这个OCR结果 写入 OCR 替换 并启动 去掉无字框
这样就 没事 临时解决方案

That’s what I did, but sometimes it recognizes differently and I can’t process all the options like that, plus the load on the PC will drop if just make ignore areas

[Mistermozg1]

你可以先 OCR一下 然后把这个OCR结果 写入 OCR 替换 并启动 去掉无字框 这样就 没事 临时解决方案

but the frames only remove what's around the text, and in the original image the text is larger than my font, so there's still ugly framing and chunks from the new grout layer

[Mistermozg1]

你可以先 OCR一下 然后把这个OCR结果 写入 OCR 替换 并启动 去掉无字框
这样就 没事 临时解决方案

That’s what I did, but sometimes it recognizes differently and I can’t process all the options like that, plus the load on the PC will drop if just make ignore areas

Yes! this is the perfect option, because there will be no load on the pc, and no unnecessary waste on deepl pro. the characters go away very quickly

[lhj5426]

就想这个样子 可以期待一下 哈哈

[bropines]

就想这个样子 可以期待一下 哈哈

Hmmm. PaddleOCR and RapidOCR recognize English and Chinese quite well, but they suck at Japanese...
Chinese:

English:

Japanese:

[lhj5426]

就想这个样子 可以期待一下 哈哈

Hmmm. PaddleOCR and RapidOCR recognize_ English _和 Chinese quite well, but they suck at Japanese... Chinese: English: Japanese:

"I mean, can we expect this feature to be added to BallonTranslator instead of using Umi-OCR? The vertical text recognition of Umi-OCR is very poor."