EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC 4.0

Real-Time Vehicle Monitoring: A Unified Framework for Detection, Tracking, and Behavioural Classification

Z. Hussain Jumah, A. M. Chyaid

AFRICAN JOURNAL OF APPLIED RESEARCH · 28.04.2026 · 12 (3) · 193-220

DOI: 10.26437/mqagpa72 ↗

Kurz gefasst

Fachartikel in der Zeitschrift AFRICAN JOURNAL OF APPLIED RESEARCH, online erschienen am 28.04.2026. Verfasst von Z. Hussain Jumah und 1 weiteren Autor:innen. Die Arbeit nennt 60 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Purpose

This paper proposes a unified framework that integrates YOLOv8s for accurate object detection and classification, DeepSORT for robust multi-object tracking, and an attention-based LSTM model for analysing temporal vehicle behaviours in urban environments. Design/Methodology/Approach: The proposed framework was evaluated using the UAVDT dataset through a structured methodology. Initially, YOLOv8s was trained to detect and classify vehicles using appropriate preprocessing and training configurations. Subsequently, DeepSORT was employed to associate detected objects across frames and maintain consistent tracking identities. Temporal features extracted from object trajectories were then fed into the LSTM-Attention model to recognise vehicle behaviour patterns. Research Limitations: The system’s performance may be affected by class imbalance in the dataset and challenges in recognising transitional or ambiguous behaviours in highly complex traffic scenarios. Additionally, deployment on resource-constrained UAV platforms requires further optimisation.

Findings

Experimental results demonstrate strong performance, achieving an overall detection precision of 89.8%, a recall of 75.3%, and an mAP@50 of 82.2%. The DeepSORT tracker achieved robust identity preservation with an IDF1 score of 87.9%, even in dense urban environments. Furthermore, the behaviour recognition module achieved an overall F1-score of 0.93, confirming the effectiveness of the proposed system across various behavioural scenarios. Practical Implications: The proposed framework can be effectively deployed in intelligent transportation systems and UAV-based monitoring platforms to enhance traffic management, improve surveillance efficiency, and support real-time decision-making. Social Implications: The system helps reduce traffic accidents by enabling early detection of risky driving behaviours and supporting smart city surveillance systems, thereby improving public safety. Originality/Value: The novelty of this work lies in integrating detection, tracking, and temporal behaviour analysis within a single unified framework, along with the use of an attention-based LSTM for improved behaviour recognition in real-world urban traffic scenarios.

Zitieren

APA

Jumah, Z. H., & Chyaid, A. M. (2026). Real-Time Vehicle Monitoring: A Unified Framework for Detection, Tracking, and Behavioural Classification. AFRICAN JOURNAL OF APPLIED RESEARCH, 12(3), 193-220. https://doi.org/10.26437/mqagpa72

Vancouver

Jumah ZH, Chyaid AM. Real-Time Vehicle Monitoring: A Unified Framework for Detection, Tracking, and Behavioural Classification. AFRICAN JOURNAL OF APPLIED RESEARCH. 2026;12(3):193-220. doi:10.26437/mqagpa72

BibTeX
@article{Jumah2026,
  author = {Jumah, Z. Hussain and Chyaid, A. M.},
  title = {Real-Time Vehicle Monitoring: A Unified Framework for Detection, Tracking, and Behavioural Classification},
  journal = {AFRICAN JOURNAL OF APPLIED RESEARCH},
  year = {2026},
  volume = {12},
  number = {3},
  pages = {193--220},
  doi = {10.26437/mqagpa72}
}
Kostenlos starten →