Added lighter variation of ML Model to make it run on the Watch. uses a text ML because CVPixelBuffer is not available as of watchOS 6.1 (Original project at: http://artizans.ai/posts/coreml-benchmark-on-iphone-11/ )
Benchmarking Deeplab's performance using Core ML on the iPhone 11 and previous devices
Most of the important code is in benchmark/ViewController.swift.
We simply load the model and the input data on initialization. When the button is pressed, we run inference several time. When the process is done, we display the number of images per seconds.
Added lighter variation of ML Model to make it run on the Watch. uses a text ML because CVPixelBuffer is not available as of watchOS 6.1 (Original project at: http://artizans.ai/posts/coreml-benchmark-on-iphone-11/ )