Convolution Input/output Units

1. What are units in CNN?
2. What is the output of convolution?
3. What is the input of a convolutional layer?
4. What should be the input size for CNN?

What are units in CNN?

Neocognitron, origin of the CNN architecture

A convolutional layer contains units whose receptive fields cover a patch of the previous layer. The weight vector (the set of adaptive parameters) of such a unit is often called a filter. Units can share filters.

What is the output of convolution?

In short, the answer is as follows: Output height = (Input height + padding height top + padding height bottom - kernel height) / (stride height) + 1. Output width = (Output width + padding width right + padding width left - kernel width) / (stride width) + 1.

What is the input of a convolutional layer?

Compared to FFNs, the early layers of a CNN allow two additional types of computation: convolution and pooling. Convolutional layers receive as input an image A^(m−1) (with K_m channels) and compute as output a new image A^(m) The output at each channel is known as a feature map, and is computed as.

What should be the input size for CNN?

The input size of each CNN is 448×448, with its initial weights transfered from the corresponding ImageNet model.