This project is a convolutional neural network (CNN for short) written in pure Java from scratch.
The projects premise is to illustrate, how and why a CNN works and operates.
The network has been trained with images from
the Stanford Dogs Dataset
and the Cat dataset as well as various other images.
As mentioned before, the project's goal is to illustrate a CNN, or AI in general. Though no fancy mechanisms are in place to graphically depict the network, it is the network and the code itself which aims to illustrate on a more abstract plane.
Instructions on building a simple CNN
Simple CNN from scratch in java
Understanding and implementing a neural network
Guide on how to construct a CNN
An introduction to CNN
Cat image dataset
Javapoint tutorial on CNNs
Convolutional Neural Networks for Visual Recognition
Github Java Guide to CNNS