Meet Remy the Cat: This devilish cat of mine (image above) inspired the concept of a smart cat door (dubbed CatNet), able to distinguish between cats in real-time and actuate a door lock accordingly.

The background story is that Remy likes to [playfully] attack another cat in the house, however this other cat does not consider Remy's actions to be playful. This presented the need for a safe room where the other cats can roam without fear of Remy lurking closely behind.

Thus, the idea for a selectively functioning cat door system -- capable of allowing other cats entrance while locking Remy out -- was born.

Deepfakes of Remy: Researching different GAN architectures to become more involved in generative modeling, Remy has turned out to be a perfect and amusing test subject.

Having gathered countless images of Remy from both the smart cat door system and a semantic (pixel-wise) image segmentation project (involving a UNet-styled CNN architecture), I now have an exceedingly robust dataset of cat images. This has yielded a snowballing effect that leaves me inclined to avoid the tedious process of data preperation and continue using my cat database for CV & ML related projects.

... Plus, it's pretty fun being able to say I have a deepfake image generator for my cat.

MLP via VHDL: The aim of this project is to create a multilayer perceptron (MLP) capable of leaning the shape of a circle and simulation it in the VHDL programming language. A MLP is one of the simplest deep neural networks for supervised machine learning applications.

Park Smart: Using computer vision and machine learning to classify the vacancy / occupancy status of individual parking spaces in real-time. This availability information is then relayed to a phone application to assist users in parking efficiency.