This program was created for Module KF_5012 - Software Engineering Practice
Group Name: Group1 / proposal_ai
Group Members: - Ryan Maclennan 17026084 Nick Reid 17006754 Ryan Fowler 17022439
This consists of two pieces of code the first one, car_classifier(.ipynb / .py) is the peice of code which creates and trains the model based off of the training data in the Stanford_Cars dataset, this code also assesses the preliminary predictions and the trained models predictions showing graphs of how the accuracy of the model has gone up throughout training and how the loss has gone down throughout training. Finally the code displays its predicitions for 30 images from the dataset colour coding the answers as green for correct and red for wrong.
The second piece of code is boundary_testing(.ipynb / .py). This piece of code loads the model from Google Drive and passes 4 individual images to the model for assessment. One image is in the dataset, one image is of a car model in the dataset but the image itself is not, one is a car but not of a make that is in the dataset and the final image is not a car but a movie screenshot to see how the system reacts when presented with such data.
Models Folder - Contains the most recent saved model Python Files Folder - Contains the .py files for the model boundary_testing_images Folder - Contains the four images for the boundary_testing code Training Images could not be uploaded to github as there are 8144 Images so this link allows access to the folder on my Google Drive : https://drive.google.com/open?id=1mkidMBTtqknef35gjrNrvmmGsVSFQMzR