Introduction

Image classification using PyTorch and Convolutional Neural Network.

How to run the program

How to run the different steps of the CNN model (and the time it takes). Paste the python commands into console/terminal.

Note: we include the console output of each stage in the stdout directory.

Step 1 - Train the network

• Train the CNN model using 350 images of 5 classes
• Save the trained model (we include pre-trained models so the file
• will show the training history of those models.)

python train_network.py

Step 2 - Test the network on the test data and report the results

• Load all the trained models
• For each model, calculate the overall accuracy of the test and training images

python dataset_accuracy.py

• Load all the trained models
• For each model, calculate the accuracy of each class
• Compute the confusion matrix
• Display the confusion matrix with Matplotlib

python class_accuracy.py