EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

An Explainable Deep Learning Framework for Crowd Density Analysis for Cause of Traffic Congestion in Makkah.

Baraa Alfasi

Computer and Decision Making: An International Journal · 03.11.2026 · 3 (6S) · 3299-3309

DOI: 10.1922/comdem.v3i6s.19068 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Computer and Decision Making: An International Journal, online erschienen am 03.11.2026. Verfasst von Baraa Alfasi. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Traffic congestion analysis has gained increasing attention recently. Transportation mobility systems are gaining popularity and are becoming more sophisticated with the problem of moving large numbers of people in terms of quantity, particularly in the area of Makkah through Hajj and Umrah. Traditional traffic analysis models are frequently unable to capture the relationships between crowd size and traffic densities. In spite of machine learning and deep learning algorithms having significant success in addressing traffic and crowd problems, including explainability, artificial intelligence can improve transparency and reveal significant parameters that explain the causes of traffic congestion. This study develops an Interpretable Deep Learning Framework for Traffic Congestion Analysis (IDLF-TCA) in Makkah. The proposed methodology employs an adaptive ReliefF-based feature selection strategy for identifying the most influential crowd-related attributes by assigning adaptive importance weights to input features. The selected features are then processed using a Tabular Prior-Data Fitted Network-based classification model for accurate identification of crowd density conditions and relevant congestion patterns. The model parameters are optimized using the AdaBelief optimizer to achieve stable convergence and improved generalization capability. Furthermore, SHAP-based explainability analysis is integrated to interpret the contribution of individual features and provide transparent insights into the factors influencing congestion conditions. The experimental analysis of the proposed IDLF-TCA framework is validated using the Hajj and Umrah Crowd Management Dataset, which supports AI and augmented reality-based applications for crowd monitoring and effective management during large-scale pilgrimages in Makkah, Saudi Arabia. The proposed framework demonstrates improved performance compared with existing methodologies, confirming its effectiveness in accurately analyzing crowd density conditions and identifying congestion patterns.

Zitieren

APA

Alfasi, B. (2026). An Explainable Deep Learning Framework for Crowd Density Analysis for Cause of Traffic Congestion in Makkah. Computer and Decision Making: An International Journal, 3(6S), 3299-3309. https://doi.org/10.1922/comdem.v3i6s.19068

Vancouver

Alfasi B. An Explainable Deep Learning Framework for Crowd Density Analysis for Cause of Traffic Congestion in Makkah. Computer and Decision Making: An International Journal. 2026;3(6S):3299-3309. doi:10.1922/comdem.v3i6s.19068

BibTeX
@article{Alfasi2026,
  author = {Alfasi, Baraa},
  title = {An Explainable Deep Learning Framework for Crowd Density Analysis for Cause of Traffic Congestion in Makkah},
  journal = {Computer and Decision Making: An International Journal},
  year = {2026},
  volume = {3},
  number = {6S},
  pages = {3299--3309},
  doi = {10.1922/comdem.v3i6s.19068}
}
Kostenlos starten →