Implementation of an Automatic Tool for Efficiently Mapping Convolutional Neural Networks to a Vector Processor Architecture