Classification exercise

[manerotoni]

Introduce the dataset without the bounding boxes. This is a little confusing
Show the labels, the different channels, and so on.

[manerotoni]

Add example that the weights should be added as tensor. torch.tensor([1,1])
You need to add the tensor to the device.
The exercise is to think about the proper weights.

[manerotoni]

save the different epochs and find the best one.
The exercise could be to load the model and run it on the whole data set