The major goal of this research is to investigate and develop computing algorithms to analyze camera and video data for intersections in CTDOT, using modern AI and computer vision technologies. The algorithms will focus on analyzing not only traffic operation information, including but not limited to vehicle/pedestrian/bicycle counts, vehicle turning movements and vehicle traveling speed, but also traffic safety metrics, including but not limited to vehicle-vehicle conflicts, vehicle-pedestrian/bicycle conflicts and unsafe intersection crossing behaviors made by pedestrians. To further extend the capabilities of investigating intersection operations and safety in more detail, these metrics will be generated by different intersection approaches, traveling directions and time periods. The team envisions that the more granular analysis can help CTDOT better optimize the signal phasing to maximize capacity and minimize delay, and identify driver behavior issues and improve safety for intersections.