High resolution and strong semantic representation are both vital for feature extraction networks of pedestrian detection. The existing high-resolution network (HRNet) has presented a promising ...