Wow, this really helps to see why they named it Tensor“Flow”. Thanks for this imagery.

Original article can be found here (source): Deep Learning on Medium

In this example, the first convolutional layer receives a size input tensor (28, 28, 1) and generates a size output (24, 24, 32), a 3D tensor containing the 32 outputs of 24×24 pixel result of computing the 32 filters on the input.