Core ML’s tools to convert other framework’s machine learning models is a pretty awesome feat in itself. However it is by no means complete. I found some issues in converting models which outputs images. Issues which took me a few weeks to solve even after many consultations to Google and Apple’s own forums. But here’s the solution and I’m sharing it with you so that you don’t have to experience the same predicament which I had.
I’ve been toying around with retraining machine learning models — one for generating prose and the other for processing photos and separate objects from its background. Here are some of the challenges I encounter as a total newbie and how I got around it.