Error (Xcode): Framework 'opencv_dart' not found

[msmartin4470]

I am sure this is me not doing something correct, but I am trying to build my iOS application using opencv_dart and I am getting an error. I tried to follow the readme as best I could and the command to download binaries seemed to work ok. Here is the error I am getting after running flutter build iOS. Any suggestions help is appreciated thanks.

Failed to build iOS app
Error (Xcode): Framework 'opencv_dart' not found

Error (Xcode): Linker command failed with exit code 1 (use -v to see invocation)

Encountered error while building for device.

[rainyl]

Are you building for real devices? If so, please refer to #24 (comment) and README in example

[msmartin4470]

Thank you. I am building for real device. I have rerun the setup command with the proper architecture, but I am still getting the same output as above. My flutter project that is using opencv_dart package is actually a plugin that my main flutter application uses. I have tried including the package and performing setup in both my plugin project and my main project.

[rainyl]

@msmartin4470 I didn't test it for a plugin, could you please give me a minimal example to reproduce it?

BTW, is it works in simulator?

[msmartin4470]

I created a base app and plugin. It has really nothing in it at all, but I seem to get the same error. I tried this with simulator also and had the same error. Thank you for your help.

https://github.com/msmartin4470/temp_app.git

[rainyl]

Confirmed, It's a bug of setup script, will be fixed quickly in v1.0.2+1

[rainyl]

v1.0.2+1 has been published, please update to v1.0.2+1 and try again, feel free to leave more message if the problem still exists.

[msmartin4470]

I believe this is working now I will do some more testing. Is your calcHist function limited to 8bit images? I do a lot of work with 16bit X-ray images and I need to use the calcHist function to get a histogram. If it is limited to 8bit images is it possible to add 16 bit support? Thanks

[rainyl]

Theoretically there is no limitation, you can use just like c++ opencv, I have tested with the following code:

final src = cv.Mat.randu(256, 256, cv.MatType.CV_16UC3, low: cv.Scalar.zeros, high: cv.Scalar.all(65535));
final mask = cv.Mat.empty();
final hist = cv.calcHist([src].cvd, [0].i32, mask, [256].i32, [0.0, 256.0].f32);
print(hist.shape);

Output:

[256, 1, 1]

if it doesn't work, please paste the error message and provide sevaral example images so I can try to fix it.

[msmartin4470]

Thank you for your help. I was getting an assertion failure, but your example works so I must have had one of my arguments wrong. Once again I really want to thank you for your help. Please keep this package up! I have been using my own custom method channel solution for about a year, but this makes things so much easier for me.

[rainyl]

Thanks~ 😄