How to reduce the package size of ML Kit iOS?

Question

I developed an application to do the text recognition by using ML Kit iOS. I only need to recognise in English. Is it possible to reduce the size of the package. Currently it's over 300 MB.

Answer 1

There are many different sizes to measure for a bundle. E.g. the bundle you upload to Apple has all 4 architectures of native code; but the actual served bundle to end user device only has 1 architecture. Another example is if you enable BitCode, the app bundle would appear much bigger locally, but the eventual downloaded size to end user device would be smaller (than wo/ Bitcode) and runs faster.

I'd recommend test for real served app size from App Store, e.g. via TestFlight.

Some numbers which I tried quickly, 1 architecture, wo/ Bitcode, 75MB for all Vision features ML Kit supports (you only need 1 of them, which is Text). 4 architectures is less than 200MB.

So some general suggestions: 1. Follow official documentation to configure only the Pods you require (instead of all Firebase). https://firebase.google.com/docs/ml-kit/ios/recognize-text#before-you-begin 2. Enable Bitcode 3. Measure real downloaded app bundle size