This was the project from CS440 Intro Artificial Intelligence. The goal was to convert greyscale images to two colorized versions of images by using K-means clustering and neural networks.

K-means clustering verion (Basic agent)

• Select 5 random pixels for grouping
• Grouping was based on the distance between the target pixel and each centroid
• Using euclidean distance to calculate the distance.

Neural Networks version (Advanced agent)

• A Feed-Forwarding 3 layers neural networks
• 9 nodes in the input layer (3x3 patch = 9 pixels)
• 5 nodes in the hidden layer
• 3 nodes in the output layer (pixel has R, G, B)
• Using sigmoid function as activation function