ImageClassifier

Implemented a convolutional neural network to classify images from the CIFAR-10 dataset.

Installation

There is no necessary libraries to run the code, except the ones included in the Jupyter Notebook. The code runs with no issues using Python 3 or newer versions.

Project Motivation

This project was part of the Data Science Udacity Program and needed to be completed in order to obtain a certificate.

File Description

This project is divided in two parts. The first part is a Jupyter Notebook, where I implemented an image classifier with PyTorch. The files involved in this part are the following:

• Image Classifier Project_Final.ipynb
• cat_to_name.json: Dictionary file. Needed for mapping the name of images.

The second part is a commmand line application for the image classifier. The files involved in this part are the following:

• train.py: Trains a new network on a dataset ans save the model as a checkpoint. Usage:

  python train.py [-h] [--save_dir SAVE_DIR] [--arch ARCH] [--learning_rate LEARNING_RATE] [--hidden_units HIDDEN_UNITS] [--epoch EPOCH] [--gpu] data_directory

  Run python train.py -h for more information.

• predict.py: Uses a trained network to predict the class for an input image. Usage:

  python predict.py [-h] [--top_k TOP_K] [--category_names CATEGORY_NAMES] [--gpu] image_path checkpoint

  Run python predict.py -h for more information.

• utils.py: Contains functions relating to the creation of the model, loading data, preprocessing images. Used by train.py and predict.py

• cat_to_name.json: Dictionary file. Needed for mapping the name of images.

Note: The two parts of this project are independent from each other.

Results

The main findings are found in the Jupyter Notebook Image Classifier Project_Final.ipynb, where markdown cells were used to walk through all the steps.