Deep-Learning-Based Aerial Image Classification for Emergency Response Applications using Unmanned Aerial Vehicles

In remote areas during emergencies and disasters, Unmanned Aerial Vehicles (UAVs) with camera sensors are essential for improving Perception of the Situation. This study Probes the use of UAVs with on-board Rooted deep learning systems for real-time Remotely piloted aerial scene Sorting in order to identify disasters like fires, floods, and collapsed buildings. We Illustrate earlier approaches and present the Aerial Image Database for Catastrophe Response (AIDER). Furthermore, we introduce a new light CNN model that Obtains a roughly threefold performance increase on embedded platforms with a small memory footprint and minimal Veracity loss (less than 2%). These results greatly advance the Review of using UAVs for real-time Emergency event recognition.