running my own vgg19 network about deep-goggle

[NerminSalem]

@aravindhm can u help?

[NerminSalem]

i tried res_inverse = invert_nn(net, y0, 'imgSize',net.meta.normalization.imageSize,'normalize',get_cnn_normalize(net.meta.normalization),'denormalize',get_cnn_denormalize(net.meta.normalization));
Error using -
Array dimensions must match for binary array op.

Error in invert_nn>nndistance (line 255)
d = x - w ;

Error in invert_nn>nndistance_forward (line 244)
res_.x = nndistance(res.x, ly.w, ly.mask) ;

Error in vl_simplenn (line 374)
res(i+1) = l.forward(l, res(i), res(i+1)) ;

Error in invert_nn (line 116)
res = vl_simplenn(net, x); % x is the random initialized image
but still the same error

[aravindhm]

I think that y0 is the wrong size. How did you set y0?

[NerminSalem]

first thanks alot for your reply
i set y0=res(end-1).x;

[aravindhm]

Unless res(end-1).x and res(end).x are the same size this will result in a size mismatch error. The network adds a l2 loss comparison at the end of net and uses y0 as the reference for this comparison.

[NerminSalem]

i will check the size of the 2 layers but if there are not of the same size, what do u recommend ?

[aravindhm]

If they are not the same size then truncate the network net.layers = net.layers(1:end-1) before passing it to invert_nn. Please let me know if this works.

[NerminSalem]

thanks alot i will try it and send u my feedback

[NerminSalem]

i m sorry where to define net.layers = net.layers(1:end-1)?

[NerminSalem]

res = vl_simplenn(net, im_) ;
y0=res(end-1).x;
net.layers = net.layers(1:end-1)

net =

layers: {1x41 cell}
  meta: [1x1 struct]

res = invert_nn(net, y0, 'imgSize',net.meta.normalization.imageSize(1:2),'normalize',get_cnn_normalize(net.meta.normalization),'denormalize',get_cnn_denormalize(net.meta.normalization),'numRepeats',1);

i m getting the following errors:
Error using vl_nnconv
The FILTERS depth does not divide the DATA depth.

Error in vl_simplenn (line 300)
res(i+1).x = vl_nnconv(res(i).x, l.weights{1}, l.weights{2}, ...

Error in invert_nn (line 116)
res = vl_simplenn(net, x); % x is the random initialized image

[aravindhm]

Something seems to have gone wrong with numRepeats == 1. I will need to investigate this. I will have time tomorrow.

[NerminSalem]

thanks alot for yr help, i really appreciate it
waiting for yr reply

[NerminSalem]

@aravindhm did u had a chance to look for the error?
thanks

[aravindhm]

I thing the commit should fix it. Use it as

res_inverse = invert_nn(net, y0, ...
'imgSize', net.meta.normalization.imageSize, ...
'normalize', get_cnn_normalize(net.meta.normalization), ...
'denormalize', get_cnn_denormalize(net.meta.normalization));

[NerminSalem]

yes it worked, thank you so much, i really appreciate it
thankssssssssssssssssssssss