Human life is invaluable, and during emergencies, rapid and accurate identification of individuals in distress is critical for effective rescue operations. Drones have become an essential tool in search and rescue missions, capturing high-resolution imagery across large and often inaccessible areas. While it's possible to identify survivors from drone footage, manual analysis is time-consuming, labour-intensive, and susceptible to error.

This deep learning model automates the detection of humans in drone imagery by drawing bounding boxes around individuals. Trained on the IPSAR and SARD datasets, the model is capable of recognizing people in a variety of complex environments, including macadam roads, quarries, tall and short grass, forested areas, and Mediterranean landscapes. Leveraging the power of deep learning, the model delivers accurate and efficient human detection, significantly reducing the time and effort required in critical search and rescue scenarios.