مجلة الاربعين المحكمة
The purpose of this study is to analyse crowd behaviour using a deep learning model. This research investigates the effectiveness of deep learning techniques in estimating crowd levels, aiming to improve accuracy and efficiency in real-time applications. The study employs a convolutional neural network (CNN), architecture trained on a diverse dataset of crowd images. Key steps in the proposed approach include data preprocessing, model training, validation, and testing. Major findings indicate that the deep learning model achieves lower error rates in both the training and testing phases, demonstrating its robustness and generalisability. Specifically, the model attained a mean absolute error (MAE) of 55.34 and a root mean squared error (RMSE) of 98.32 during testing, compared to 56.5 MAE and 99.13 RMSE during training. These results highlight the model’s capacity to generalise well to new, unseen data. The study concludes that deep learning models, when properly trained and validated, can significantly enhance crowd-level estimation, offering valuable insights for applications in public safety, event management, and urban planning. Future work will focus on expanding the dataset and refining the model to further improve performance and applicability. Overall, our study presents a promising advancement in crowd counting technology with practical implications for crowd management at large-scale events like the Arbaeen Pilgrimage.