What Is Depthwiseconv2d And Separableconv2d? How Is It Different From Normal Conv2d Layer In Keras?

I was looking through the architecture of EfficientnetB0 and noticed DepthwiseConv2D operation. Did some digging and found that there's also a SeparableConv2D. What exactly are the

Solution 1:

DepthwiseConv2d performs the first step in a depthwise spatial convolution, apply convolution operation on each input channel separately.