feat: Instance Segmentation API

[benITo47]

Description

This PR introduces Instance Segmentation Module. Provided API allows for using two predefined models (RF-DETR and YOLO26-Seg family) as well as plugging custom models through fromCustomConfig.

On top of adding new API, this PR adds common CV utilities on the CPP side, as well as migrates Object Detection to leverage them.

Introduces a breaking change?

• Yes
• No

Type of change

• Bug fix (change which fixes an issue)
• New feature (change which adds functionality)
• Documentation update (improves or adds clarity to existing documentation)
• Other (chores, tests, code style improvements etc.)

Tested on

• iOS
• Android

Testing instructions

• Run test suite
• Try new features by running demo app for Instance Segmentation
• Confirm ObjectDetection works as expected by running respective demo apps

Screenshots

Related issues

#825

• I have performed a self-review of my code
• I have commented my code, particularly in hard-to-understand areas
• I have updated the documentation accordingly
• My changes generate no new warnings

Additional notes

[msluszniak]

There are some broken links in documentation, can you fix them?

JC has time off and is out of office

[msluszniak]

I will look at this PR in a minute

[msluszniak]

Suggested change

	return cv::Size(shape[shape.size() - 2], shape[shape.size() - 1]);
	return {shape[shape.size() - 2], shape[shape.size() - 1]};

[msluszniak]

Why do we pass tuple here, these should be 3 separate function, each fetching different type of data.

[msluszniak]

Suggested change

	return cv::Rect(mx1, my1, mx2 - mx1, my2 - my1);
	return {mx1, my1, mx2 - mx1, my2 - my1};

[msluszniak]

Suggested change

	return cv::Rect(x1, y1, x2 - x1, y2 - y1);
	return {x1, y1, x2 - x1, y2 - y1};

[msluszniak]

These tests are not triggered by the script test

also I get some errors which doesn't occured earlier:

/Users/msluszniak/test_bare_rn/react-native-executorch/packages/react-native-executorch/common/rnexecutorch/tests/integration/TextToSpeechTest.cpp:77: Failure
Expected: Kokoro(kValidLang, kValidTaggerPath, kValidPhonemizerPath, kValidDurationPath, kValidSynthesizerPath, "nonexistent_voice.bin", nullptr) throws an exception of type RnExecutorchError.
Actual: it throws std::invalid_argument with description "File not found: kokoro_en_tagger.json".

[ FAILED ] TTSCtorTests.InvalidVoicePathThrows (0 ms)

This one was fixed with LLM & TTS integration

/Users/msluszniak/test_bare_rn/react-native-executorch/packages/react-native-executorch/common/rnexecutorch/tests/integration/BaseModelTests.h:121: Failure
Type parameterized test suite CommonModelTest is defined via REGISTER_TYPED_TEST_SUITE_P, but never instantiated via INSTANTIATE_TYPED_TEST_SUITE_P. None of the test cases will run.

Ideally, TYPED_TEST_P definitions should only ever be included as part of binaries that intend to use them. (As opposed to, for example, being placed in a library that may be linked in to get other utilities.)

To suppress this error for this test suite, insert the following line (in a non-header) in the namespace it is defined in:

GTEST_ALLOW_UNINSTANTIATED_PARAMETERIZED_TEST(CommonModelTest);

[ FAILED ] GoogleTestVerification.UninstantiatedTypeParameterizedTestSuite (0 ms)

/Users/msluszniak/test_bare_rn/react-native-executorch/packages/react-native-executorch/common/rnexecutorch/tests/integration/BaseModelTests.h:108: Failure
Expected: Traits::callGenerate(model) doesn't throw an exception.
Actual: it throws rnexecutorch::RnExecutorchError with description "The model's forward function did not succeed. Ensure the model input is correct.".

/Users/msluszniak/test_bare_rn/react-native-executorch/packages/react-native-executorch/common/rnexecutorch/tests/integration/BaseModelTests.h:109: Failure
Expected: Traits::callGenerate(model) doesn't throw an exception.
Actual: it throws rnexecutorch::RnExecutorchError with description "The model's forward function did not succeed. Ensure the model input is correct.".

/Users/msluszniak/test_bare_rn/react-native-executorch/packages/react-native-executorch/common/rnexecutorch/tests/integration/BaseModelTests.h:110: Failure
Expected: Traits::callGenerate(model) doesn't throw an exception.
Actual: it throws rnexecutorch::RnExecutorchError with description "The model's forward function did not succeed. Ensure the model input is correct.".

[ FAILED ] TextToImage/CommonModelTest/0.MultipleGeneratesWork, where TypeParam = rnexecutorch::models::text_to_image::TextToImage (4981 ms)

[msluszniak]

There are some thing to fix in demo app:

1. Why there is object detection live when there is a separate vision camera screen, also when clicking on this, I got unmatched route.
2. Why instance and semantic segmentation are not next to each other on the list below
3. There is no way to get back from vision camera example, I missed it in previous review, cc: @NorbertKlockiewicz

[NorbertKlockiewicz]

3. You can just swipe from the left side

[msluszniak]

@NorbertKlockiewicz I don't thing that is good enough. If I had a problem and haven't thought about that swipe, then potentially many more users might have the same problem (there is no intuition that there might be something here). That's my opinion.

[msluszniak]

Also we should fix the format how tags are aligned to the bounding box. Below there is an example of it shouldn't look like. If you don't want to fix this altogether with this PR, please make a separate issue please.