Search and rescue of maritime distress personnel is a frontier topic of world research. Aiming at the problem of search and rescue for traditional maritime distress personnel, a mathematical model of search and rescue for maritime distress personnel is proposed, which can be carried on unmanned aerial vehicles. A search and rescue system for maritime distress personnel is established by using YOLO (you only look once) real-time target detection algorithm. The system can send UAV(Unmanned Aerial Vehicle) with high-definition camera equipment to search for people in distress after the ship sends out distress and abandoned ship rescue signal, and make a further decision on the image information returned by UAV and YOLO algorithm. The experimental results and feasibility analysis show that the system can process the video image information returned by UAV better, and the information can be used for the decision-making of rescue personnel.