CODEWITHSUNDEEP
tensorflowconcatenationconv-neural-networkconvolution
Devin Haslam
Devin Haslam

Reputation: 759

concatenating convolutions with different strides in tensorflow

I am attempting to reproduce a CNN from a research paper using tensoflow. Here is the whole architecture of the CNN, but I am mainly focused on the Reduction A section.Whole Architecture

Reduction A

I am wondering if I have spotted a problem with the research paper. As you can see in Reduction A, 3 layers are concatenated. However, 2 of those layers use a stride of 2. Therefore, when concatenating the tensor along the 4th axis(number of channels), the right most layer does not have the same depth, width and height as the other 2 layers. I am aware that I could use padding to fix this, but there is no mention of this in the paper. Do you believe this research paper has a mistake? Should the right most path of reduction A also use a stride of 2?

Upvotes: 1

Views: 659

Answers (1)

BlueSun
BlueSun

Reputation: 3570

Considering that all the other reductions and inceptions have matching strides, it seems like the paper made a mistake. I suppose the 3x3(384) convolution was supposed to have a stride of 2, since this convolution increases the channel size.

Upvotes: 2

Related Questions