FACE-DETECT: AUTOMATIC FACIAL DETECTION IN CHALLENGE SCENARIOS

Abstract

Automatic face detection is a key technology for applications involving security, intelligent surveillance, access control, and environmental monitoring. However, factors such as facial occlusions, illumination variations, scale changes, pose diversity, and uncontrolled conditions remain significant challenges for computer vision systems. This paper presents Face-Detect, a deep learning-based approach for automatic face detection in challenging scenarios. The proposed methodology was evaluated using the Face Obstruction Detection, WIDER FACE, and AI Face Dataset benchmarks, covering situations involving masks, glasses, multiple faces, and synthetic images. Experiments were conducted using YOLOv8 and YOLOv8n-face architectures, enabling a comparative analysis of different detection strategies. The results demonstrated satisfactory levels of precision, robustness, and generalization capability, confirming the feasibility of the proposed approach for face detection in challenging scenarios. The analyses performed demonstrate the potential of the solution for applications in intelligent surveillance, public safety, and automated monitoring systems.