Tensorflow Lite errors when running in iOS devices

[cmalbuquerque]

Running my application in iOS device (iPhone 7 with iOS 14.4) it crashes when the model is processing the data. I believe that happens due to Tensorflow Lite Errors (see output) but I have no idea how to fix it:

carolinaalbuquerque ~/Documents/audio_recognition_app (*main) > flutter run
Launching lib/main.dart on iPhone de Carolina in debug mode...
Automatically signing iOS for device deployment using specified development team in Xcode project: 5SSNTW7HP4
Running Xcode build...                                                  
 └─Compiling, linking and signing...                        19,5s
Xcode build done.                                           28,9s
(lldb) 2021-02-20 18:18:18.623713+0000 Runner[414:13363] Warning: Unable to create restoration in progress marker file
fopen failed for data file: errno = 2 (No such file or directory)       
Errors found! Invalidating cache...                                     
fopen failed for data file: errno = 2 (No such file or directory)       
Errors found! Invalidating cache...                                     
Installing and launching...                                        36,1s
Initialized TensorFlow Lite runtime.
TensorFlow Lite Error: Regular TensorFlow ops are not supported by this interpreter. Make sure you apply/link the Flex delegate before inference.
TensorFlow Lite Error: Node number 2 (FlexSize) failed to prepare.

Failed to create the interpreter with error: Failed to allocate memory for input tensors.
["0 Background Noise", "1 Bell", "2 Whistle", "3 Xylophone"]
Activating Dart DevTools...                                         5,9s
Syncing files to device iPhone de Carolina...                       176ms

Flutter run key commands.
r Hot reload. 🔥🔥🔥
R Hot restart.
h Repeat this help message.
d Detach (terminate "flutter run" but leave application running).
c Clear the screen
q Quit (terminate the application on the device).
An Observatory debugger and profiler on iPhone de Carolina is available at: http://127.0.0.1:53066/z1fkaZhV7VE=/

Flutter DevTools, a Flutter debugger and profiler, on iPhone de Carolina is available at:
http://127.0.0.1:9101?uri=http%3A%2F%2F127.0.0.1%3A53066%2Fz1fkaZhV7VE%3D%2F

Running with unsound null safety
For more information see https://dart.dev/null-safety/unsound-null-safety
requesting permission
start microphone
recordingBuffer length: 11008
recordingBuffer length: 22016
recordingBuffer length: 33024
recordingBuffer length: 44032
reached threshold
Running model
* thread #21, queue = 'conversionQueue', stop reason = EXC_BAD_ACCESS (code=1, address=0x0)
    frame #0: 0x00000001d6d0a128 libsystem_platform.dylib`_platform_memmove + 72
libsystem_platform.dylib`_platform_memmove:
->  0x1d6d0a128 <+72>: stnp   x12, x13, [x0]
    0x1d6d0a12c <+76>: stnp   x14, x15, [x0, #0x10]
    0x1d6d0a130 <+80>: subs   x2, x2, #0x40             ; =0x40 
    0x1d6d0a134 <+84>: b.ls   0x1d6d0a158               ; <+120>
Target 0: (Runner) stopped.
Lost connection to device.

I am using a Google Teachable Machine model and I followed these steps for iOS configuration.

I have tested already in Android device and it works perfectly but I need to guarantee iOS support!

[Caldarie]

Hi @cmalbuquerque

the error that sticks out the most is: TensorFlow Lite Error: Regular TensorFlow ops are not supported by this interpreter. Make sure you apply/link the Flex delegate before inference.

Usually, you will get this error if you run it on an emulator, but it seems like you are using a device.

The most likely possibility is that your select ops haven’t been configured correctly. Have you force load it? Instructions here

[cmalbuquerque]

@Caldarie yes, I forced load as mentioned on documentation:

And after specify the force load, I ran into iOS folder:

$ flutter pub get
$ pod install
$ flutter clean

[Caldarie]

To check whether it’s a bug or not, can you load your model and labels on the example project provided in this repository? Can you tell me if you receive the same error?

[cmalbuquerque]

Yes @Caldarie, I tested using the example project and I had the same error... Also, I checked your iOS configurations for GTM models and are the same than mine. 😞

[Caldarie]

If it’s ok with you, would you be willing to share your model with me? It’s possible that google may of changed something, or the people responsible for iOS tensorflow made a breaking change.

[cmalbuquerque]

Yes, of course! Maybe you're right and some update from them breaks something...
Here is my basic model: gtm_model.zip Thanks! 😄

[Caldarie]

Hi @cmalbuquerque,

I was able to run your model fine (as shown on the image below).

There are few possibilities for your problem:

1. It's possible that your editor (vscode or xcode) has failed to detect your device and you are running the project on an emulator. To detect the device, you need to run a few steps: Instructions here

2. The other possibility is that your configuration may be incorrect. But i think that is least likely considering that you ran it with my example

3. Another possibility is that you may have forgotten to add the following in your pubsec.yaml. However, i also think this is also very unlikely.

   - assets/soundclassifier.tflite
   - assets/labels.txt

[cmalbuquerque]

Thanks! Could you tell me what's your Xcode version?

I reviewed all steps of my configuration and I noticed that when I run flutter pub get into /ios directory, it changes automatically the ios/Flutter/Release.xcconfig file to:

#include? "Pods/Target Support Files/Pods-Runner/Pods-Runner.release.xcconfig"
#include "Pods/Target Support Files/Pods-Runner/Pods-Runner.release.xcconfig"
#include "Generated.xcconfig"

And running pod install:

carolinaalbuquerque ~/Documents/audio_recognition_app/ios (*main) > pod install    
Analyzing dependencies
Downloading dependencies
Generating Pods project
Integrating client project
Pod installation complete! There are 3 dependencies from the Podfile and 5 total pods installed.

[!] CocoaPods did not set the base configuration of your project because your project already has 
a custom config set. In order for CocoaPods integration to work at all, please either set the base 
configurations of the target `Runner` to `Target Support Files/Pods-Runner/Pods-Runner.profile.xcconfig` 
or include the `Target Support Files/Pods-Runner/Pods-Runner.profile.xcconfig` in your build configuration
(`Flutter/Release.xcconfig`).

Also, if I replace my ios/Flutter/Release.xcconfig file with yours, the same warning appears. The suggestion given by warning has already in file, so I don't understand what might be misconfigured 😞 Here is my Xcode configurations:

[Caldarie]

Hmm, I’ll get back to you about this problem

As for Xcode version, I’m using 12.4

[cmalbuquerque]

@Caldarie thanks for your time! I tried to reinstall Xcode, update pods and finally it works! Sorry for your time and thanks again! 😁

[Caldarie]

Glad I could be of assistance!