OCR Manga Reader for Android

EPWING support and a way to correct OCR errors. 'Nuff said. The correction interface isn't as good as Kaku by 0xbad1d3a5, but then again Kaku doesn't have EPWING support. Most errors come from either 1) you messing up with your selection (or the auto-selection functionality not helping you enough), 2) hurigana problems, 3) bad scans, or 4) common errors that are to some extent fixable if you update your substitutions.txt. Mine is below if you're interested. Nothing special. It IS admittedly a bit annoying though to sometimes see simple kana or kanji be misinterpreted as something completely ridiculous and uncommon, even when there is no hurigana in the way. I don't know how feasible it is, but I wonder if there could be a setting to select a different OCR result or to default to a more common character even if its probability of being a match is slightly lower (perhaps within some range you can configure, which the user might want to adjust based on what they're reading). Anki integration is roundabout but alright. Not sure it can get much better on Android, frankly - which is why I use the x86 version on an emulator on PC. Glad that exists. Dev is great. Release cycles are a bit long, but the fact that he is still alive and even remotely responsive to bug reports is good enough for me. I've never developed for Android, and according to the developer of Kaku, at least, "Not entirely sure how to build OCR Manga Reader. I think I tried to do it a couple years ago then gave up. Pretty sure it's also missing some JNI bindings in the source code for Leptonica so you won't be able to simply "just build it" even on Android Studio. My memory is fuzzy though so I might be wrong." This sort of deters me from digging around the source code myself. # This is the OCR Manga Reader OCR substitutions file. ぁ=あ ぃ=い ぅ=う ぇ=え ぉ=お カゝ=か ヵ=カ ゎ=わ 〈=く <=く ＜=く ︿=く \|二=に _=一 I=ー 1=ー ,= = \.= L=し 『= -=ー ~=ー 一=ー

Best OCR manga reader for android for sure, just has some annoying quirks like being very inconsistent in detecting small tsu, which seems to stem from the fact that each character is cropped individually instead of finding an optimal character bounding box size for all detected characters. It also struggles sometimes to detect characters in a way that suggests that images are downsampled too much before running the actual OCR.

This is without a doubt the best OCR translation tool available on android and, as absurd as it sounds, It's completely free. The layout is well designed, the controls are very intuitive and there are plenty cool features of customization options available. It took me just a few minutes to learn to use this app properly. The compatibility with EPWING dictionaries is by far my favourite feature and that's where my only concerns come from: unlike the edict dictionary included in the app, whenever I use an Epwing dictionary only the results found for the maximum (1st pic) or minimum (2nd) number of kanji are displayed (in the first case it was no big deal but in the second one it made a big difference since I was using enamdict in Epwing format). So I was wondering if it was possible to either include enamdict in the app or have an option to force the Epwing dictionaries to behave like the included edict when needed; it would also be cool to have a button or a shortcut (for instance, a left/right swipe) to cycle through the dictionaries. Thank you so much for your hard work and sorry for making requests in the middle of a review. i58.tinypic.com/osev6p.png i62.tinypic.com/16c97xh.jpg EDIT: You are awesome! I just tried the optional enamdict dictionary bundled with v4 and now everything works perfectly! I wish I could award you with 6 stars!