Plot_top_losses() clipping input data warning

fastai
1

Hi all,
When I run:

interp = SegmentationInterpretation.from_learner(learn)
interp.plot_top_losses(4)

I get this warning:

WARNING:matplotlib.image:Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).

and the resulting images have different colors.

I read that one possible solution can be to multiply the output :
plt.imshow((out * 255).astype(np.uint8))

How could I integrate it with Fastai?

1 Like