The College of Computer Science and Information Technology at University of Basrah organized a scientific lecture entitled “Multi-Object Tracking Based on Artificial Intelligence Algorithms.” Researcher Batoul Baji Aboud presented the key findings of her master's thesis on multi-object tracking using artificial intelligence techniques.
The lecture aimed to provide a concise overview of the problem of tracking multiple pedestrians in videos, particularly in crowded scenes where people overlap, their paths intersect, or they are partially obscured. This is one of the most significant challenges facing intelligent tracking systems.
The lecture included a presentation of the Selective ByteTrack Fuzzy system, which represents an improvement on the ByteTrack algorithm's linking mechanism while retaining the YOLOX detector unmodified. The system handles clear cases according to the original mechanism, while dealing with ambiguous cases using a Mamdani-type fuzzy inference system based on five factors: spatial overlap, detection confidence level, motion consistency, path reliability, and uncertainty resulting from obscuring.
