How to remove the adaptive average pool layer from vgg19?

Question

I have loaded the pre-trained model of vgg19. How to remove the adaptive average pool layer which is present before the classifier?

@mrzo In pytorch i would like to know how i should be doing this? — i_am_somebody, Dec 10 '19 at 16:51

score 3 · Answer 1 · answered Dec 10 '19 at 22:56

3

If you are using PyTorch and torchVision model, you can disable the use of the last maxpool layer like this -

nn.Sequential(*list(vgg.features._modules.values())[:-1])

Here all the layers are in an array, and not including the last element does the trick.

answered Dec 10 '19 at 22:56

Solaiman Salvi

thanks. but, doing this also removes the max pooling layer. how to retain the max pooling and remove only the adaptive avg pooling layer? – Naveen Reddy Marthala Aug 14 '22 at 07:38
and how to drop only the Dense layers at the end retaining both max and adaptive-avg pool layers? – Naveen Reddy Marthala Aug 14 '22 at 07:38
my torch version is `1.12.1+cu113` and torchvision's is `0.13.1+cu113` on python 3.7.13 on Ubuntu 18.04. – Naveen Reddy Marthala Aug 14 '22 at 07:38

1 Answers1