Effective Face Detector Based on YOLOv5 and Superresolution Reconstruction

Table 4

Performance comparison using different models.

Model	Backbone	AP50	Time/ms
HR	Resnet101	57.5%	198
YOLOv3	Darknet53	57.9%	51
Ours	YOLOv5s-SRGAN	59.8%	75