Learning OpenCV 4 Computer Vision with Python 3 - Third Edition

Choosing the size of the input layer

The number of nodes in the input layer is, by definition, the number of inputs into the network. For example, let's say you want to create an ANN to help you determine an animal's species based on measurements of its physical attributes. In principle, we can choose any measurable attributes. If we choose to classify animals based on weight, length, and number of teeth, that is a set of three attributes, and thus our network needs to contain three input nodes.

Are these three input nodes an adequate basis for species classification? Well, in a real-life problem, surely not – but in a toy problem, it depends on the output we are trying to achieve, and this is our next consideration.