sv.plot_image using inches?!?

[AlainPilon]

I just realized plot_image size argument uses inches, with a default or 12x12.

Other than making it easy to pass it to the underlying math.pyplot.figure, is there a specific reason to use inches? This is highly impractical for anyone working in a Jupyter notebook as the image resizes based on the window size.

My suggestion would be to use a pixel based size with a default value using the original image size.

If for some reason we need to keep the size in inches (!), at least update the doc to replace:

size (Tuple[int, int]): The size of the plot.

with

size (Tuple[int, int]): The size of the plot in inches.

[LinasKo]

It indeed is a thin wrapper around pyplot, and that's the only reason inches are used. We don't want to drift from it quite yet.

I'll push a fix to clarify the docs, and add this to the list of smaller issues to look at in the future.

[AlainPilon]

Thanks.

I had my working window rendering the picture ALMOST the same size as the original and it took me way too much time figuring out why my annotations didnt line up.

[SkalskiP]

You can always convert the image to Pillow and display it in collab. Just use cv2_to_pillow. I believe Pillow displays images in original resolution.