Post AanexAwbQCmm86aEEa by julianfairfax@fosstodon.org
 (DIR) More posts by julianfairfax@fosstodon.org
 (DIR) Post #Aanex9S4yCm9V9IL1E by julianfairfax@fosstodon.org
       2023-10-15T10:52:00Z
       
       0 likes, 0 repeats
       
       I have an Android phone at the moment, but plan on switching to a Linux phone in the future. However, unfortunately not everything is available on it, even through websites.For example, DeepL's website has no support for translating images, yet its proprietary mobile app (which is also obviously only available on iOS and Android), does.How do you mobile Linux users workaround these kinds of issues? How would you translate an image using your mobile Linux device, without writing all the text?
       
 (DIR) Post #AanexAAkI8VLjgC0dk by cjr@fosstodon.org
       2023-10-15T13:08:12Z
       
       0 likes, 0 repeats
       
       @julianfairfax First a disclaimer: I don't have a Linux phone (yet), so if anyone can test these apps on mobile device, that would be great.That being said, one can use Frog to extract Text from Images, and Dialect to translate the text.Both apps are for GNOME :gnome: and are available as Flatpaks .For more information:Frog: https://github.com/tenderowl/frogDialect: https://github.com/dialect-app/dialect/
       
 (DIR) Post #AanexAwbQCmm86aEEa by julianfairfax@fosstodon.org
       2023-10-15T15:55:09Z
       
       0 likes, 0 repeats
       
       @cjr I tried Frog on my PinePhone, using an image taken with Megapixels. So, admittedly, the quality it's working with is not high at all.Regardless, the result was quite awful. Aside from the words not being recognised correctly, which might be a quality issue, Frog doesn't stick lines from the same sentence together. That makes it not very useful for translation.But all this aside, maybe Dialect could add support for OCR itself? Maybe by learning from Frog? @rafaelmardojai
       
 (DIR) Post #AanexBfGk8VyMdTtr6 by rafaelmardojai@social.odka.org
       2023-10-15T16:37:58Z
       
       0 likes, 0 repeats
       
       @julianfairfax @cjr Yes, image translation within Dialect is on the roadmap, but it is not a high priority since Frog exists.One of the main blockers is the implementation of the models/download manager, which we also need to implement for offline translation providers.
       
 (DIR) Post #AanmxNhQnTftK561L6 by xerz@fedi.xerz.one
       2023-10-15T18:07:46.847514Z
       
       0 likes, 0 repeats
       
       @julianfairfax @cjr @rafaelmardojai Frog just uses whatever Google's Tesseract gives it, with no further processing ― that's the real issue.