NavBLIP: a visual-language model for enhancing unmanned aerial vehicles navigation and object detection by Ye Li, Li Yang, Meifang Yang, Fei Yan, Tonghua Liu, Chensi Guo, Rufeng Chen

“…Furthermore, NavBLIP employs a multimodal control strategy that dynamically selects context-specific features to optimize real-time performance, ensuring efficiency in high-stakes operations.Results and discussionExtensive experiments on benchmark datasets such as RefCOCO, CC12M, and Openlmages reveal that NavBLIP outperforms existing state-of-the-art models in terms of accuracy, recall, and computational efficiency. …”
Get full text