Could you explain what label_np is? Are these the logits from your model or is it the target?
hi @ptrblck
this is the target which I am converting to numpy array . Sorry for the bad naming convention 
It’s fine. I was just wondering, why you have float values in your target, but it’s because you are interpolating, right?
Running your code produces a color image on my machine. Maybe your preprocessing has a minor bug?
This is the code I’m running:
label_np = np.random.randn(1, 3, 24, 24)
label_np[label_np<0]=0
label_np[label_np>3]=2
label_np[np.logical_and(label_np>0.6,label_np<=3)]=1
label_np[np.logical_and(label_np>0,label_np<=0.6)]=0
color1 = color_transform(torch.from_numpy(label_np))
temp_image1 = image_transform(color1)
temp_image1.save("temp_image_00001.png")
hi ,
yes … Float value is coming for interpolation.