feat: use MobileNetV4 Core ML model with NCHW input

[anhappdev]

The models can be downloaded from here: https://drive.google.com/drive/folders/1-Mloub0e41mkYs09i9tQZl95dTnGdZ6f?usp=sharing

[github-actions[bot]]

MLCommons CLA bot All contributors have signed the MLCommons CLA ✍️ ✅

[anhappdev]

There are 2 issues:

1. I cannot get a plausible accuracy. I am not sure what is wrong with the code?
2. There is no official way to know which format the input data has or the benchmark_id to know when to trigger mlperf_backend_convert_inputs()?

[freedomtan]

@freedomtan to check the NHWC -> NCHW conversion.

[freedomtan]

@anhappdev this NHWC -> NCHW looks right to me. Maybe you can try to compare with the result of using a single transpose op TFLite model (Yes, it's posssible. E.g., in TFLite label_image, I constructed a TFLite model with only resize op to resize the input image to the size I needed, https://github.com/tensorflow/tensorflow/blob/master/tensorflow/lite/examples/label_image/bitmap_helpers_impl.h#L30-L100)

[sonarqubecloud[bot]]

Quality Gate passed

Issues
3 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarCloud

[anhappdev]

There was an issue with wrong data type, but I got it working now.