Developing Facial Recognition-Based Reconnaissance to Enhance Stadium Security and Detect Banned Fans at Soccer Stadiums

Soccer or football outside of North America remains a major global industry but faces ongoing issues with hooliganism and stadium violence, which often involves banned fans. Despite the severity of these challenges, research on stadium security is limited with most studies utilizing Convolutional Neural Networks (CNNs) for tasks such as attendance and emotion detection rather than crowd security. This research introduces a facial recognition system aimed at identifying banned soccer fans to enhance stadium security. This system leverages Siamese CNNs for accurate facial detection and matching using publicly available data. Data preprocessing and augmentation strategies improved the model resilience, which leads to a detection accuracy of 94.45% for unoccluded faces and 89.85% for occluded faces as it surpasses traditional CNN models by a significant margin. Additional testing addressed demographic bias and robustness to occlusions, which results in enhanced fairness and balanced accuracy across different racial groups. This research offers a reliable and unbiased facial recognition-based reconnaissance solution tailored to soccer stadium security. The findings can guide the development of best practices for the deployment of facial recognition technologies in sports venues by contributing to safer and more secure environments.