watson-developer-cloud / visual-recognition-coreml

Classify images offline using Watson Visual Recognition and Core ML
Apache License 2.0
490 stars 77 forks source link

crop image to center before classification #31

Closed devinaconley closed 6 years ago

devinaconley commented 6 years ago
bourdakos1 commented 6 years ago

I don’t know if we should crop the photos to square before classifying. Most people don’t crop their photos before training so the results may suffer. I know the training data that you made was cropped to a square though. How should we handle this?

devinaconley commented 6 years ago

merging now - will update later on making this clear to users, both at train time and run time