Brennan was a MASc student who worked on novel motion-aware strategies for efficient and accurate video analytics. His thesis highlights the significant potential of leveraging motion information in video analytics, particularly in the domains of pose estimation and action recognition. It introduces a novel approach, POOF (Pose annotation using Optical Flow), that remarkably reduces the annotation burden for pose estimation models while substantially improving accuracy. Additionally, the thesis presents M2A (Motion-Aware Attention), a breakthrough in action recognition, showcasing substantial accuracy enhancements by effectively incorporating motion information into attention-based mechanisms, making these models more efficient for real-world applications.