Date of Award
5-2026
Culminating Project Type
Starred Paper
Styleguide
ieee
Degree Name
Computer Science: M.S.
Department
Computer Science and Information Technology
College
School of Science and Engineering
First Advisor
Jalal Khalil
Second Advisor
Jie H Meichsner
Third Advisor
Bhaskar Ghosh
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.
Abstract
Multi-camera vehicle re-identification (Re-ID) aims to associate the same vehicle across multiple non-overlapping traffic cameras, enabling applications in traffic monitoring and intelligent transportation systems. While prior research has demonstrated strong performance on curated benchmark datasets, comparatively less attention has been given to corridor-based deployments using publicly available traffic camera feeds without ground-truth identity labels. This study presents a staged vehicle re-identification approach for traffic surveillance that integrates YOLOv8 detection, multi-object tracking, and OSNet-based appearance embeddings to associate vehicle tracklets across sequential cameras. Cross-camera associations are performed using cosine similarity combined with temporal feasibility constraints and tiered appearance consistency filtering. Evaluation on MnDOT traffic camera footage using manual verification produced 71.4% precision for pairwise associations and 50.0% precision for four-camera trajectories across 131 associations, with performance decreasing as trajectory length increases. Results indicate that most errors arise from appearance ambiguity under constrained resolution rather than temporal infeasibility, suggesting that staged corridor-based association is feasible for short-range vehicle tracking but requires stronger appearance representations for longer trajectories.
Recommended Citation
Waqar, Taha, "Multi-Camera Vehicle Re-Identification Using YOLOv8 Detection, ByteTrack Tracking, and OSNet Appearance Embeddings" (2026). Culminating Projects in Computer Science and Information Technology. 74.
https://repository.stcloudstate.edu/csit_etds/74
