Doorknob State Synthesizer and Network-Based Location Monitoring

As an optional personal project that aligned with my course study, I investigated the capabilities of human location tracking via network monitoring and utilizing computer vision in a home surveillance application. On a single host system, computer vision algorithms were used to analyze doorknobs and deadbolts in order to synthesize their locked/unlocked state while home network traffic was constantly analyzed to deduce what residents are currently home and the times that residents come/leave home. This collective information allows the owner of the system to know if there house was left unlocked and unattended in addition to which resident left the home in that state.


An Implementation of K-Means Image Segmentation on Massively Parallel GPU Architecture using CUDA

In this project, I recreated a K-Means Image Segmentation approach in CUDA. The kernel scheme for this approach was inspired by similar work from a research group from the University of Split, Croatia.