<p>Detecting people with the help of unmanned aerial vehicles (drones) is becoming indispensable in search and rescue operations, disaster relief, monitoring public or sporting events, and similar activities to increase people's safety. However, there are still many challenges today, such as inaccurate sensors on the drone that affect the quality of the collected data, movements, and variability of the camera and recording system set up under the influence of weather conditions, complex scenes with small objects of interest, etc. In this paper, a fine-tuned YOLOv8 model is proposed for real-time person detection on images captured by a monocular camera on a drone, and three geolocation algorithms for people detected in search images retrospectively based on metadata collected during the drone flight. The proposed passive geolocation algorithms calculate the possible GPS coordinates of a person detected in an image based on an ellipsoidal model of the Earth and then with an extension that includes a digital elevation model based on data from two image locations. The proposed person detection and geolocation algorithms were extensively tested with two commercial drones in different terrains and under different conditions of movement/location of the drone and the person being searched for. Excellent results were obtained, 91% mAP for person detection and about 1–10&#xa0;m error in determining geolocation in different terrain and movement/location conditions. Based on the results obtained, we can recommend the use of the proposed model and geolocation algorithms in search and rescue operations to increase their success while reducing the required resources.</p>