Visualize images in intermediate layers in a torch (lua)

Question

In the conv-nets model, I know how to visualize filters, we can do itorch.image (model: get (1) .weight)

But how could I effectively render output images after convolution? especially these images in the second or third layer in a deep neural network?

Thanks.

Answer 1

Similar to weight you can use:

itorch.image(model:get(1).output) 

Answer 2

To visualize the scales:

 -- visualizing weights n = nn.SpatialConvolution(1,64,16,16) itorch.image(n.weight) 

To visualize function maps:

 -- initialize a simple conv layer n = nn.SpatialConvolution(1,16,12,12) -- push lena through net :) res = n:forward(image.rgb2y(image.lena())) -- res here is a 16x501x501 volume. We view it now as 16 separate sheets of size 1x501x501 using the :view function res = res:view(res:size(1), 1, res:size(2), res:size(3)) itorch.image(res) 

More details: https://github.com/torch/tutorials/blob/master/1_get_started.ipynb