Cloud-Based Image Segmentation Approach for Internet of Drones (IoD)

We propose a novel cloud-based image segmentation framework tailored for the Internet of Drones (IoD), leveraging advanced neural network architectures to process aerial imagery. Our approach utilizes a modified U-Net model, implemented at a cloud-enabled Base Station, to ensure efficient, scalable image analysis without burdening drone resources. The system capitalizes on high-speed 5G connectivity for real-time data transmission, achieving significant improvements in segmentation accuracy as demonstrated on a diverse 24-class dataset. Our results indicate a promising enhancement in computational efficiency and a reduction in drone energy consumption. This research contributes to IoD applications in smart city planning, agriculture, and surveillance by optimizing the synergy between drone technology and cloud computing capabilities.