مجلة الاربعين المحكمة
The Arbaeen pilgrimage is one of the largest human gatherings in the world, posing significant challenges for traffic management and traffic sign supervision due to extreme population density, frequent changes in temporary traffic routes, and the physical and mental fatigue experienced by organizational staff. These factors make it increasingly difficult to guide visitors, particularly the elderly and those unfamiliar with the area. This research aims to address these challenges by developing an intelligent, automated system capable of instantly recognizing Arabic traffic signs and providing audio alerts to drivers and traffic coordinators. A convolutional neural network (CNN) model was employed, incorporating data analysis techniques such as preprocessing, data augmentation (including rotation and zooming), and normalization. The system was trained and tested on a publicly available dataset (ArTS Dataset) containing Arabic traffic signs under diverse conditions (e.g., lighting, angle, clarity). An interactive graphical user interface (GUI) was designed, allowing users to upload images of traffic signs (simulating camera input) and receive real-time predictions accompanied by audio pronunciation in Arabic. The model achieved test accuracy exceeding 90% when evaluated on the designated test dataset, along with strong performance in additional metrics such as precision, recall, and F1-score. Although the data used does not originate from actual traffic signs in Karbala during the Arbaeen pilgrimage, it reflects similar real-world conditions. This work represents an initial step toward future field deployment, particularly in smart transportation systems, where such a system could support visitors in their native language and in real time, thereby enhancing public safety and mobility during large-scale religious events.