Cyclist Detection, Tracking, and Trajectory Analysis in Urban Traffic Video Data

The major objective of this thesis work is examining computer vision and machine learning detection methods, tracking algorithms and trajectory analysis for cyclists in traffic video data and developing an efficient system for cyclist counting. Due to the growing number of cyclist accidents on urban roads, methods for collecting information on cyclists are of significant importance to the Department of Transportation. The collected information provides insights into solving critical problems related to transportation planning, implementing safety countermeasures, and managing traffic flow efficiently. Intelligent Transportation System (ITS) employs automated tools to collect traffic information from traffic video data. In comparison to other road users, such as cars and pedestrians, the automated cyclist data collection is relatively a new research area. In this work, a vision-based method for gathering cyclist count data at intersections and road segments is developed. First, we develop methodology for an efficient detection and tracking of cyclists. The combination of classification features along with motion based properties are evaluated to detect cyclists in the test video data. A Convolutional Neural Network (CNN) based detector called You Only Look Once (YOLO) is implemented to increase the detection accuracy. In the next step, the detection results are fed into a tracker which is implemented based on the Kernelized Correlation Filters (KCF) which in cooperation with the bipartite graph matching algorithm allows to track multiple cyclists, concurrently. Then, a trajectory rebuilding method and a trajectory comparison model are applied to refine the accuracy of tracking and counting. The trajectory comparison is performed based on semantic similarity approach. The proposed counting method is the first cyclist counting method that has the ability to count cyclists under different movement patterns. The trajectory data obtained can be further utilized for cyclist behavioral modeling and safety analysis