The aim was to classify the RGB colored images into the given 10 classes. The dataset is the CIFAR-10 dataset. The original link to the dataset is provided below from where you can download the dataset: https://www.cs.toronto.edu/~kriz/cifar.html
As explained in the link above, the dataset is balanced with each classes having 6000 images and a total number of 10 classes. The training set is 50000 and test set is 10000.
Methods Implemented: ANN with one and two hidden layers and CNN. Moreover, the original images were also converted into grayscale and classified to see the difference in the accuracies between the colored and grayscaled version of the dataset.
The main.m file contains all the methods implemented. The grayscaled and original images are also available as mat files.
Note: Grayscaled images and test and training images were stored as .mat files and called directly via the filename in the main.m.