👉 Depthwise is a type of neural network layer in computer vision and machine learning. It is used to reduce the spatial dimensions of an input image, often by convolving it with a kernel that has a larger filter size than the input. The depthwise operation involves two convolutional layers (usually 2D) performing a horizontal or vertical convolution on the input image, which then maps each pixel in the output to its corresponding location within the original image. This process is repeated for each of the spatial
