Welcome to my website!

On this site you will find two interactive showcase projects presenting different Machine Learning models and capabilities.

The first is a Semantic Segmentation Building Identification project, which allows you to search for any location on Earth using the ubiquitous Google Maps JavaScript API, and send it to a Google Cloud Platform Cloud-Run instance, where a custom-trained Detectron2 CPU-model will attempt to locate any buildings in the image, and send it back to be displayed as an overlay.

The second is a Google Quick, Draw! based interactive and comparative project. This project presents two separately trained models using two different resolutions (28x28 pixels and 56x56 pixels) to attempt to identify your drawing from one of 345 discrete classes of objects. It also shows a representation of 'what the model sees' by way of a visualizing the conversion performed of your drawing before being sent to the model.