Adding FastViT & MobileViT image classification models

[veb-101]

Hi, I've recently finished implementing Apple's FastVit , MobileVit, and MobileViT-v2 image classification models in my repo. I've also ported their weights from the original models. My only pending task for these is making them channel-first compatible, but that's simple. I'm also planning to least train the FastViT models on Pascal VOC and make it available, as there is a lack of model options in TensorFlow for object detection since the closing of TFOD.

Could these models be added to Keras-Hub as well? I couldn't find a Vision-specific contributing guide, or do I need to follow either resnet or mobilenet structure?

My final question is whether these models are needed here? If you guys need support for implementing some other models, you can let me know.

[veb-101]

One more thing, the licences on these models is this: github link
Will it be a problem?