Search
Wuhan University and Ant Group have collaboratively introduced a groundbreaking semantic-enhanced remote sensing model named SkySense++. Leveraging a comprehensive dataset comprising 27 million multimodal remote sensing images, this model utilizes a novel two-stage progressive learning approach, markedly enhancing the efficacy and practical application of remote sensing data processing. The research team rigorously tested SkySense++ on twelve Earth observation tasks spanning seven key domains, including agriculture, forestry, and disaster management. The model exhibited exceptional performance, especially in classification, detection, and segmentation tasks, achieving substantial improvements over current methodologies. Furthermore, the latest iteration boasts the ability to deploy without fine-tuning, thereby further augmenting the model's generalizability and ease of use.
-
C114 Communication Network -
Communication Home
