Why the conv1 layer has padding 2?

ishay2b / VanillaCNN

Implementation of the Vanilla CNN described in the paper: Yue Wu and Tal Hassner, "Facial Landmark Detection with Tweaked Convolutional Neural Networks", arXiv preprint arXiv:1511.04031, 12 Nov. 2015. See project page for more information about this project. http://www.openu.ac.il/home/hassner/projects/tcnn_landmarks/ Written by Ishay Tubi : ishay2b [at] gmail [dot] com https://www.l

188 stars 80 forks source link

Why the conv1 layer has padding 2? #16

Closed huinsysu closed 6 years ago

huinsysu commented 6 years ago

Hi, in the vanilla_train.prototxt why the conv1 layer has the parameter pad 2? If do so, how can we get the size of 16 18 18 map after the first convolution layer and polling layer?

ishay2b commented 6 years ago

The code was written on a slightly different version of the article network architecture (different padding), so map size is different.