manga-ocr VS scenic

I use manga-ocr on pc

I usually read manga, and use yomichan and manga_ocr. Initially I'll try to read the sentence by itself to see if I understand it. If I don't, I'll take a screenshot of the sentence so manga_ocr parses the text, and then I'll paste it somewhere in my browser and use yomichan to check unknown vocabulary. I've already done Genki I and II so I almost never have to look up grammar. When I do find grammar I don't understand, a quick Google search, or just referring back to Tae Kim or Genki, will do. I never translate the full sentence, I just check the meaning of unknown words and try to understand the sentence in Japanese.

If on browser, use Yomichan. Otherwise, set up a screenshot tool like ShareX with https://github.com/kha-white/manga-ocr

manga-ocr, which you will probably want to use for convenience anyway, convert the text to horizontal and it will automatically show up this way in the Yomichan clipboard monitor.

MangaOCR on github is really good. There are two main gui of it that i know of. The first is a whole gui reader of it and is called poricom. The second lets you ocr anything on the screen when pressing alt+q and is called cloe. Here are the links : https://github.com/kha-white/manga-ocr https://github.com/blueaxis/Poricom https://github.com/blueaxis/Cloe

I'm going to go against what everybody is saying and say to not rely on furigana. There was another comment saying to set up mokuro to use yomichan with. Imo, that would be the more ideal setup. You not only have access to a lot more manga (that don't use furigana), but being overly reliant on furigana may prevent you from building up your intuition for figuring out which kanji reading to use in which moment. Part of building comprehension/intuition is trying to figure out which kanji reading/word meaning you should use in each differing context. Furthermore, if you can't set up mokuro, look into setting up an OCR and a texthooker. https://github.com/kha-white/manga-ocr The one above would be the one that I would recommend the most.

they are probably referring to this Great thing, reads an image from the clipboard and puts back the text from the image.

If you have more of these, manga-ocr can read it without any issues.

Anyone figured out how to run this against a video?

https://github.com/google-research/scenic/tree/main/scenic/p... has an example showing how to "train Vid2Seq on YouCook2" using "python -m scenic.projects.vid2seq.main", but I couldn't see the recipe for using it against a video to return a description.

There are a few libraries/frameworks that one can use and allow to reuse the same code for datasets, logging, training loop etc.... . E.g. Lightning or Scenic. Maybe you can use one of these or at least get some inspiration for your own code.

Quick Read: https://www.marktechpost.com/2022/11/25/google-research-proposes-an-artificial-intelligence-ai-model-to-utilize-vision-transformers-on-videos/ Paper: https://openaccess.thecvf.com/content/ICCV2021/papers/Arnab\_ViViT\_A\_Video\_Vision\_Transformer\_ICCV\_2021\_paper.pdf Github link: https://github.com/google-research/scenic/tree/main/scenic/projects/vivit

GitHub: https://github.com/google-research/scenic

The SCENIC code, etc., has been open-sourced on the project’s GitHub. The paper SCENIC: A JAX Library for Computer Vision Research and Beyond is on arXiv.