NASA And IBM AI Models Improved With Global Data Training

One of the data used to train Prithvi Geospatial (photo: dock. NASA)

JAKARTA Last year, NASA along with IBM and German research institute Forschungszentrum Jölich launched a basic model of Artificial Intelligence (AI) named Prithvi Geospatial Open-source.

When introduced for the first time, NASA said that this AI fundamental model could support various geographic applications. Now, with global data participation, the AI model Prithvi can do more, including helping farmers.

NASA explained that the AI basic model Prithvi could help farmers track changes to the land used, could monitor possible disasters and whether the land would be disrupted, as well as predictions of worldwide harvests.

"We have instilled NASA's scientific expertise directly into this basic model, which allows them to quickly convert data petabytes into follow-up insights," said NASA's Chief Science Data Officer Kevin Murphy.

Kevin added that this change made Prithvi Geospatial feel like a great assistant. By using this AI model, users can make decisions faster and more precise so that economic and social benefits can be generated.

Before becoming a sophisticated tool, this basic model was originally trained with a large data set, along with self-learning techniques. After that, researchers used HLS satellite imagery that could represent various landscapes and avoid low-quality data caused by clouds or weather.

After several trials and training continues to be improved, basic Geospatial model Prithvi has proven to be able to work well in several applications. The role of this basic model is even very important for a number of Earth monitoring applications.

Prithvi Geospatial can map post-disaster floods and detect burns caused by fires. Some of the applications known to use this AI-based model are Multi-Tempural Cloud Gap Imputation and Multi-Temporal Plant Segmentation.

"This refined model has undergone rigorous testing in various use cases, ensuring increased flexibility and performance, producing a model version that empowers environmental monitoring applications, and provides significant social benefits," said Rahul Ramachandran, Base Model AI leader at Marshall.

Simak breaking news dan berita pilihan Anda langsung di ponsel. Pilih saluran andalan akses berita voi.id WhatsApp Channel : https://whatsapp.com/channel/0029VamlU850gcfBaMTjry0w. Pastikan Anda sudah install aplikasi WhatsApp.