Abstract: Based on artificial intelligence and visual control technology,a progressive target focusing method for UAV head based on visual control is proposed,which firstly uses deep learning to build a dual detection model to obtain the visual characteristics of power equipment components,and secondly uses a multi-stage progressive algorithm to adjust the camera focus and control the head to rotate and shift to autonomously capture the target of attention and achieve center focus.Quantitative experiments show that this method can focus,identify and analyze component defects of UAV aerial photography in real time and effectively improve detection efficiency and accuracy,which provides reference and enlightenment for problems such as multi-target and small-scale detection of complex electric power components and defects.