The use of Unmanned Aerial Vehicles (UAVs)equipped with classification systems in Search and Rescue missions has increasingly shown great application potential. There are however pratical concerns regarding limited embedded computing power and adapting the classifier performance to the characteristics of a specific mission. The UAVs may acquire a large number of images during a mission and it is difficult for humans to search throughout this material. For this matter, we study the performance of a leaner Convolutional Neural Network model to select images with salient objects and present a tool that can help in decision making regarding the definition of the classification threshold through the analysis of possible operational impacts and consequences of eventual detection errors. We also show that, in this application, Data Augmentation is important to improve the classifier performance.

Traditional maritime Search and Rescue (SAR) operations face notable limitations, including reliance on manual observation and the high costs associated with deploying search vessels and aircraft, which often contribute to delayed response times. In contrast, the advent of Unmanned Aerial Vehicles (UAVs) equipped with high-resolution sensors has demonstrated considerable potential in enhancing SAR efforts. UAVs provide rapid aerial surveillance over expansive areas, offering a cost-effective, flexible solution with a lower environmental footprint. The recognition of the benefits these technologies can bring to SAR missions has driven researchers to investigate computer vision based object detection methodologies. Ongoing advancements in machine learning and computer vision are focused on creating robust object detection systems that can operate in real-time, even in the dynamic and challenging conditions typical of maritime environments. This paper investigates integrating computer vision techniques into UAV-based maritime SAR operations. It offers a comprehensive review of current research, including types of sensors, datasets, and object detection methods, along with the evaluation metrics used to assess system performance. By analyzing the current state of this field, the study aims to identify key challenges, propose viable solutions, and outline future research directions to further advance UAV-based SAR capabilities.