Question regarding depth convolution
venuiitg opened this issue · comments
Hi Sayak Paul, Thank you for the TF converted models. I have noticed that you have used conv2D with groups to perform a depth convolution.
I am wondering if you had tried using https://www.tensorflow.org/api_docs/python/tf/keras/layers/DepthwiseConv2D instead? Just checking if there is a reason for preferring one over the other.