COMP 551 - Mini-Project 3: Image Classification

The second project for COMP 551 - Applied Machine Learning.

This repository contains various algorithms used to classify images into forty categories. Using the InceptionV3 model, we achieved a score of 89.152% accuracy on an unseen testing data.

Dependencies

Before running any of the models, install:

• numpy
• tensorflow
• pandas
• opencv
• keras

Running the models

To run the Keras model, run:

python cnn_pretrain.py

To run the InceptionV3 model, run:

python tensorFlow.py

To run the Logistic Regression model, run:

python LogisticRegression.py

To run the Neural Network model, run:

python nn_cross_val.py

To test the Neural Network model on scikit-learn's digits dataset run

python digits_cross_val.py

Included directories

The output file for the logistic regression model are found in the output folder. The results for the InceptionV3 model are directly generated by the tensorFlow program.