The use of drones in search-and-rescue missions allows us to easily search areas that are inaccessible to humans and enables rapid and efficient mission execution with minimal manpower. In this paper, we propose a search operation method that involves automatically recognizing missing persons based on real-time images captured by a camera mounted on a drone and estimating their geolocation information. Given a particular search area, we plan a flight path while taking into consideration a cost function with constraints. Using a deep-learning model trained using cropped, generated, and augmented data, we recognize missing persons through real-time images taken by the drone following the planned path. Additionally, we estimate the geolocation of the missing persons by coordinate-transforming the reference pixels of recognized objects in the image. Based on the estimated geolocation, we identify identical objects and count the total number of objects recognized during missions. We validate the proposed search method by completing a search-and-rescue challenge using a drone.