Hi guys my second Android application- Photo OCR Image to text converter.I made it in 5 days,hence it doesn't look very good

Question

https://play.google.com/store/apps/details?id=awesomeocr.leehar.com.awesomeimage2textconverter. It works correctly in my Moto g4 device, but still has bugs in some devices like in Samsung galaxy J2 ,the picture taken using camera doesn't get added to gallery.Also in an old Samsung device running KitKat,after capturing, camera app stops without displaying image.Even in my device, photo captured goes into camera folder.I followed the Android docs and it is supposed to store in separate directory.

Answer 1

will getExternalStorageDirectory automatically create the app directoy or should I also include the code to make directory.the code doesn't mention that. I'm confused

Answer 2

OCR is a tough beast.
First things first:
I'm on an Honor6 with Android KitKat 4.4.2
the app background coud use some rework as it feels out of contest.
ads are reasonable and well placed (do not interfere too much with the app)
Camera crashed before saving the photo.
as expected text in screenshots is well detected while photos of printed text give erratic results.

in case of scanning tables the output seems a bit jumbled
and it may be wise to have a shortcut to the scanned picture for quicj reference.
the app works well offline: thumbs up for avoiding online ocr services.
What OCR engine did you use?

Answer 3

thank you seamiki. I used Google vision library. I too tried it on a KitKat device and it crashed,but photo was saved into the gallery. I am trying to solve that bug. I intend to make layouts for different screen sizes.Oh you meant the photo of that girl,yes I will replace that with a new relevant photo.googles vision library is good. I tried it by taking the photo of a page of a book I was reading and it extracted all the test.but 1 or 2 sentences at the end turned up at the beginning. I think the photo has to be captured as straight as possible to avoid this error.And you are right screenshots work better. I learnt machine learning a while back and thought of making an ocr scanner from scratch, but then I realized as of now I only needed to include the google library in project which explains the large number of ocr apps available. I am actually trying to get job as an Android developer, so I actually want to make one more app which recognizes and changes hair color in a photo to include in my cv.I don't think google will help me with that.I will have to do the alg on my own.Thank you so much for checking out my app and reviewing it over here.

Answer 4

had some headaches with jtesseract some time ago. The post editing of the picture was the real royal pain (contrast, bw, format conversion etc...) and that ended me up in using a backend program (capture2text) to do the ocr (ugly hack but it got the job done and saved me a lot of time).

Answer 5

cool😉👍so in effect as time progresses more libraries will come up which will make the job of developers easier but will in turn lead to more apps consequently leading to less market potential for an app idea.like even if somebody comes up with a better ocr than Google's it will be very hard to get it noticed among all these other apps, I guess