Image Classification Model

How It Works

Image Classification Model

How It Works

Image classification is one of the many applications of artificial intelligence. An image classification model does exactly what it sounds like: it classifies images. While this may seem simple for a human, for a computer, an image is just a bunch of 1s and 0s. However, once a computer is able to identify an object, it can do so much faster than a human ever could.

I recently created my own image classification model to identify characters from Star Wars. Check out the video I made here!

Image classification has a huge number of use cases. Image classification algorithms are more accurate and faster than experts who trained in their fields for years. Image classification could be used for sorting images on your phone, identifying tumours, and even improving the accuracy of self-driving cars!

How it Works

Transfer Learning

Transfer learning is the process of training a model to complete a task it was not originally trained to do. I used transfer learning to retrain the Inception v3 model to identify Darth Vader, Count Dooku, Darth Maul, Kylo Ren, and Darth Sidious.

Image Classification

When looking at a picture, a human can easily identify objects. Computers, on the other hand, see numbers. Every pixel of an image is just its RGB value, so it is really complicated for a computer to see things like lines, shapes, edges, etc.

Convolutional Neural Networks

A convolutional neural network, or CNN, works by going over data over and over, in a process known as a convolution. This is very useful in image classification as convoluting over an image over and over lets a computer identify distinct features of an object.

Contact Me