Improving the accuracy of face detection under varying sizes and lighting conditions using multiscale learning methods

This paper proposes a face recognition method based on multi-scale learning, aimed at improving the accuracy of face detection under varying sizes and lighting conditions. Through the use of Feature Pyramid Networks (FPN) and attention mechanisms, the model is capable of better capturing and recognizing faces of different sizes and proportions. Experimental results demonstrate that this method achieves excellent performance on multiple public face detection datasets.