Image discolouring

[miriskalt]

Hey guys!

So after hearing the first lesson, I tried to replicate the code given. Herefore I use colab.
After training the model I displayed the top losses with the interp.plot_top_losses(9, figsize=(15,11)) command. Yet the images looked as follows:

Dubble checking if the images looked like in the lesson before I accepted the mystery. Yet, I set up my own data training set and noticed the same discolouring.
Since I didn’t change any code from the lesson1 notebook I won’t upload the code here again.

Can someone offer a tip, as to why this is happening? The learning rate etc. was not influenced by this. Also the images looked normal when I run the data.show_batch(rows=3, figsize=(7,8)) command after the training:

Thanks in advance for your help!
Kind regards.

[dhoa (Dien-Hoa)]

This the the Grad-CAM heat map. We use it to understand which part of the image is more important for the decision. Details you can find here: https://docs.fast.ai/vision.learner.html

To remove it, just pass heatmap = False:
plot_top_losses ( k , largest = True , figsize = (12, 12) , heatmap : bool = None , heatmap_thresh : int = 16 , return_fig : bool = None ) → Optional [ Figure ]

Hope that helps,

[miriskalt]

Ah, they are colourful again.

Thank you very much for that easy fix!
Seems legit now.