To make it possible to control an image capturing position and its direction in water easily and flexibly through the use of a miniature unmanned aerial vehicle equipped with a plurality of rotors. This is solved by an underwater image capturing apparatus including a miniature unmanned aerial vehicle equipped with a plurality of rotors, a winding machine capable of delivering and winding a string-like member, and an underwater camera capable of capturing images in water, wherein the winding machine is fixed to the miniature unmanned aerial vehicle and the string-like member is connected to the underwater camera.