Image Classification Data

Not sure I understand your question.

Could these tutorials help you?

https://developer.apple.com/documentation/createml/creating_an_image_classifier_model

or

h t t p s : / / w w w .raywenderlich.com/7960296-core-ml-and-vision-tutorial-on-device-training-on-ios

Otrherwise, please explain more what is your problem.

The answer depends on how you want to classify images and why.

The first step in the process is that someone, somewhere, looks at an image and assigns a label (Classification) to it. The same image can have different labels, depending on purpose (the "why"): an image of a red Lamborghini can be classified as "Car", "Red vehicle", "Luxury sports-car" etc. Somewhere on the web there will probably be sets of images (with labels) for common classifications, such as vehicle types (car, truck, bus). However, if your purpose is to identify cars with cracked windscreens you'll probably have to create your own images and then apply labels yourself: "Not cracked", "Slightly cracked", "Badly cracked".

The labelled images, whatever their source, are then used to train (create) a model using CreateML and evaluate it based on pre-assigned labels. Thereafter the output model can be used to determine the classification of any new image presented to it.

For common tasks, such as determining if an image contains a face (person), there are already established models - which may or may not be publicly available, at a price or free. Such models might be compatible with CreateML and/or CoreML, directly or via some other process, in which case there's no need for you (someone) to go through the classification/training process yourself.

Regards, Michaela