Solar photovoltaic power generation is an important component of a country's energy structural adjustment. With the rapid expansion of the scale of the photovoltaic power generation industry in recent years, the need for an automated daily maintenance of photovoltaic power stations has increased. Traditional manual detection methods are inefficient because photovoltaic power stations are spread over a large area. In this study, we investigate the intelligent inspection technology of a photovoltaic power station based on an unmanned aerial vehicle (UAV). A technical route for an intelligent inspection of a UAV-based photovoltaic power station is proposed. We achieve the automation of photovoltaic panel image data acquisition and analysis and investigate defect detection based on computer vision. We realize photovoltaic panel defect detection based on infrared images using an adaptive dynamic threshold method combined with image enhancement technology, facilitating the classification of defects to be determined by using visible light images. The defect locations are further calculated by combining the POS data and the camera model. Finally, we verify the effectiveness of the proposed technical route in an actual scenario.