1/ PSA/rant: If your'e a developer trying to port a deep learning model from Onnx to Core ML and are finding that Apple's onnx-coreml conversion script isn't working out of the box- you aren't alone. While iOS13 has brought about great improvements to what you can do in Core ML,
-
Show this thread
-
/7 Also also.... Just put all your image normalization/denormalization code in additional layers sandwiching your model before converting it to Onnx. It's easier to reason about and do correctly where as Apple's code again is strange and not quite right.
2 replies 0 retweets 7 likesShow this thread -
Replying to @citnaj
I agree that it's not seamless but: 1/ it's way better than it used to be. 2/ the onnx conversion itself is not seamless either, in my experience 3/ You get amazing perf on device in the end, which for me makes all of this worth it 1b/
@bhushan23s lead on onnx-coreml is v helpful1 reply 0 retweets 2 likes -
PS: this thread really makes me want to try DeOldify on an iOS device!! Do you already have a CoreML app out?
1 reply 0 retweets 1 like
Working on it ha
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.