Machine Learning Home Work II Venice boat classification for the Master of Engineering in Computer Science
The objective of the project is to build an image classifier with CNN using only a few training examples - just a few hundred or thousands of images from each class of different categories of boats navigating the City of Venice (Italy). In other words, given an input image, with a computer automatically classify it into one from a set of categories, say “Gondola”, “raccolta rifutti” or “ambulanza”, etc.
The dataset used here is the MarDCT from the Sapienza University of Rome. The training dataset contains 4.774 images divided into 24 classes. The test/validation dataset contains 1.969 images, and Ground truth text file which contains images divided into their respective classes.