Research on the use of Deep Learning for Mask Object Recognition Under Different Lighting Conditions

In the current environment where technology is developing rapidly and real-world scenarios are becoming increasingly complex, the application scenarios where precise recognition of individuals wearing various types of face coverings under different lighting conditions are becoming more and more common, especially in areas such as public safety, intelligent access control, and personalized services, which is also the purpose of this study. When the detected person is wearing a mask, the YOLOv11 algorithm can only recognize the mask-wearing situation, while the Xception model can recognize the person's expression, but neither can recognize the person's identity. Therefore, this paper proposes a method that combines the deep learning algorithms of FaceNet and RetinaFace to obtain the five key facial feature points of a person, and then performs database matching to obtain the identity result of the detected person.