CarletonU Building Classification

Assignment 3 for SYSC4906. The goal is to build an image classification model to recognize a subset of buildings on Carleton Univeristy campus.

The approach taken is a convolutional neural network. Training makes use of concepts such as data augmentation and dropout. The end results was a model that predicted new images from a randomly selected holdout set with 86% accuracy.

Contents

1. trainModel.ipynb contains all the code to build and train the model from samples images
2. deployModel.ipynb contains all the code to test the model with new images
3. model.pkl is a serialization version of the model trained model
4. SYSC4906_Assig3_final.ipynb is the contents of both of the previous notebooks combined
5. SYSC4906 Assignment 3.pdf is the orginal handout for the assignment

the archive file containing all the images in the dataset is unfortuantely too large to upload here