Implementation of artificial neural networks using NumPy in addition to extraction of features and classification of the Fruits360 image dataset
This project builds artificial neural network in Python using NumPy from scratch in order to do an image classification application for the Fruits360 dataset. Just 4 classes are used from such a dataset which are Apple Braeburn, Lemon Meyer, Mango, and Raspberry.
At first, features are extracted from the dataset using the extract_features.py script. This file is expected to be located at a directory in which there are 4 folders holding the images of the 4 classes. The folders are named apple, lemon, mango, and raspberry. The script loops through the images within the 4 folders for calcualting the features which are the color histogram of the hue channel of the HSV color space. The script saves 2 files. The first one holds the features of all samples and the second one is the class labels of the samples.
After preparing the training data (inputs features and class labels), next is to implement the ANN and train it according to such data. This is done using the ann_numpy.py script. This project is documented in a tutorial titled Artificial Neural Network Implementation using NumPy and Classification of the Fruits360 Image Dataset and available at my LinkedIn profile here: https://www.linkedin.com/pulse/artificial-neural-network-implementation-using-numpy-fruits360-gad. This tutorial just implemented the forward pass of the ANN without implementing the backward pass for updating the ANN paremters (i.e. weights).
There is an extension to the previous tutorial in order to use the genetic algorithm (GA) for optimizing the network weights which increases the classification accuracy. The tutorial is titled Artificial Neural Networks Optimization using Genetic Algorithm and available at my LinkedIn profile here: https://www.linkedin.com/pulse/artificial-neural-networks-optimization-using-genetic-ahmed-gad. The GitHub project implementing the second tutorial is available at my GitHub page here: https://github.com/ahmedfgad/NeuralGenetic.
Everything (i.e. images and source codes) used in both tutorial2, rather than the color Fruits360 images, are exclusive rights for my book cited as "Ahmed Fawzy Gad 'Practical Computer Vision Applications Using Deep Learning with CNNs'. Dec. 2018, Apress, 978-1-4842-4167-7 ". The book is available at Springer at this link: https://springer.com/us/book/9781484241660.
The source code used in this tutorial is originally published in my GitHub page here: https://github.com/ahmedfgad/NumPyANN
For contacting the author
Towards Data Science: https://towardsdatascience.com/@ahmedfgad