This project focuses on utilizing Cross Attention Networks (CAN) for the detection of Remotely Operated Underwater Vehicles (ROVs). By exploring various feature modalities, the CAN model enhances the detection capabilities of ROVs in challenging underwater environments. This method improves accuracy by incorporating multi-dimensioNol data from different sensors, such as image, thermal, and soNor inputs. Through cross attention, the network intelligently aligns features from these different modalities, improving the overall detection performance and operatioNol reliability of ROVs for missions such as underwater exploration, infrastructure inspection, and environmental monitoring.
Accessibility Controls