China has made new progress in the infrastructure construction of large models in vertical fields. Recently, Huantian Smart Technology Co., Ltd. officially released the selection announcement for the bidding agent of the "Satellite Remote Sensing Large Model Application Training Pilot Base Project," marking that the project's preparatory work has fully entered the substantive advancement phase. According to the announcement, the estimated total investment of the project reaches 60 million yuan.

As a key project focusing on the integration of cutting-edge AI and space-aerial information, this pilot base project has set clear and ambitious technical development goals. In terms of core models and data, the project plans to develop a satellite remote sensing large model with parameters exceeding 10 billion, while building a satellite data sample library of more than 100 million scales, thus providing solid data support for the deep training of large models.

In terms of specific function development, the project will focus on overcoming two key capabilities: first, developing AI interpretation capabilities for more than 30 types of satellite data, and second, achieving accurate identification of more than 20 land cover categories. These technological achievements will directly empower the entire industry ecosystem of remote sensing satellite applications.

To comprehensively promote the optimization upgrade and technology transformation of the remote sensing large model, the base will systematically build seven core functional modules, including scenario factory, model school, corpus base, computing power engine, verification and evaluation, talent origin, and technology transformation, as comprehensive training bases. Through these supporting facilities, the project will achieve efficient training optimization and iterative upgrades of the remote sensing large model.

In addition, this pilot base will also provide professional artificial intelligence large model pilot platform services for more than five practical application scenarios such as agricultural remote sensing, further accelerating the deep integration of space-aerial information and the real economy.