Ant Group launches SkySense, a 2 billion parameter multi-modal remote sensing basic model

According to news from this site on February 28, Ant Group announced the launch of the 2 billion parameter multi-modal remote sensing basic model SkySense, which supports text, infrared light, visible light, SAR radar and other modes, and supports multi-resolution Time-series remote sensing image modeling can be used for landform, crop observation and interpretation, etc., effectively assisting agricultural production and management. According to reports, the R&D staff Ant Bailing’s large model multi-modal capabilities used an internally constructed 1.9 billion remote sensing image data set for pre-training, and finally obtained the 2.06 billion parameter model SkySense. Officials stated that this is the largest multi-modal remote sensing model with the largest parameter scale, the most complete coverage of tasks, and the highest recognition accuracy in the world so far.

This site noticed that the official information shows that SkySense ranked first in the above 17 evaluations and can be widely used in Data and identification services are currently provided through Ant's internal MEarth platform in important areas such as urban planning, forest protection, emergency disaster relief, green finance, and agricultural monitoring.

The above is the detailed content of Ant Group launches SkySense, a 2 billion parameter multi-modal remote sensing basic model. For more information, please follow other related articles on the PHP Chinese website!