Also known as: Depthwise Separable Convolutional Neural Network (sepCNN) Β· sepCNN Β· Depthwise Separable Convolutional Neural Network sepCNN
Depthwise separable convolutional neural network (sepCNN) is a convolutional architecture that replaces standard convolutions with more efficient depthwise separable convolutions. These operations separate spatial filtering from channel mixing, reducing computation while preserving useful representation capacity.
A convolutional neural network architecture based on Inception, but where Inception modules are replaced with depthwise separable convolutions. Also known as Xception. A depthwise separable convolution (also abbreviated as separable convolution) factors a standard 3D convolution into two separate convolution operations that are more computationally efficient: first, a depthwise convolution, with a depth of 1 (n n 1), and then second, a pointwise convolution, with length and width of 1 (1 1 n). To learn more, see Xception: Deep Learning with Depthwise Separable Convolutions.