missemlizb / annotate-ml Goto Github PK

Cf "Training Object Detection Models in Create ML" (2019) at 3:06

Great Job! I like this app. Tried to create a data set for Turi Create and found an issue.
Exported files have all coordinates scaled which shrinks and moved them relative to origin.
In tracing through your code, in AddPhotos in Document.swift:

 guard let data = image.tiffRepresentation,
					let image = NSImage(data: data) else {
					continue
				}

It takes an image file which both Preview and Photoshop is, say 1024 x 800, and the resulting image from the NSImage call is about a third the size(when original resolution is 240 dpi). The scale factor is different for each input image. I suspect it has to do with differing image resolution among images.

Hi, @MissEmlizB

I met this error when select training data in CreateML Object Detection trainer.

"Empty table from specified data source."

I used Annotate-ML ver 1.1.

What's wrong to me ?

Thanks in advance.
Best,
@bemoregt.