[ML Kit Bug] On Device Text Recognition Crash about mlkit HOT 6 CLOSED

I'm unable to reproduce this issue on Firebase 6.4.0. Can you try updating and seeing if the crash is fixed?

from mlkit.

Tested today on the latest Firebase available (6.5.0). I see with profiler that there is some memory leak, which consumes memory up to 9.88 Gb and then crashes. Tested on iphone 6 iOS10.3.2. Takes about 13 minutes to reproduce. Starting from minute 8 the screen is like slideshow already.

from mlkit.

Does your profiler show the leak is coming from the Firebase object? Trying to understand how to consume 10Gb from a single image on device text detection. If you were doing live camera, I could've seen images building up.

from mlkit.

I can't say exactly where the leak is. I'm not in touch with iOS development, so i've just run this sample. See the screenshot attached from the profiler. It is default xCode Instruments.

It is not a single image, I selected 'live' recognition. So processed frames from camera preview 'on fly' during 13 minutes until memory reached critical value.

from mlkit.

Probably I've selected the wrong thread. So let me know if I need to create another issue.

from mlkit.

Hey @enriquemgz, we've made a number of improvements to our samples with the newest ml kit release, and we believe this should be fixed now. If you're still seeing the issue, please do let us know!

from mlkit.