Photovoltaic power station light-shielding inspection method based on unmanned aerial vehicle and deep learning image recognition

The invention discloses a photovoltaic power station light-shielding inspection method based on an unmanned aerial vehicle and deep learning image recognition, which comprises an unmanned aerial vehicle with a shooting function and an upper computer capable of performing man-machine interaction with the unmanned aerial vehicle, and is characterized in that the upper computer issues a command to the unmanned aerial vehicle according to whether a light spot area exists or not on an acquired image, and when the light spot area does not exist, the unmanned aerial vehicle performs normal inspection; when the light spot area exists, the unmanned aerial vehicle rotates by a corresponding angle according to the illumination direction until no light spot exists in the collected and fed-back image, the unmanned aerial vehicle stops rotating by the angle, image collection is carried out at the angle, and inspection is carried out. The method effectively eliminates the interference of the light spot on the image shot by