（全球TMT2023年8月8日讯）IBM与Hugging Face合作宣布，他们基于NASA卫星数据构建的IBM watsonx.ai地理空间基础模型已经在Hugging Face平台上发布。这是Hugging Face平台上规模最大的地理空间基础模型，也是首个与NASA合作构建的开源AI基础模型
NASA与IBM合作发布了Hugging Face平台上最大的开源地理空间AI基础模型
IBM and NASA collaborated to train this foundational model, utilizing Harmonized Landsat Sentinel-2 (HLS) satellite data from the past year within the continental United States. The model was fine-tuned based on annotated data of flood and burn areas. Compared to current leading technologies, this model achieved a 15% improvement in performance using only half the labeled data under equivalent conditions. With further refinement, the model can also be applied to new tasks such as tracking deforestation, predicting crop yields, and detecting and monitoring greenhouse gases. Researchers from IBM, NASA, and Clark University also partnered to utilize this model in fields like time-series segmentation and similarity studies.
以上是Hugging Face发布了基于NASA卫星数据构建的AI地理空间基础模型的详细内容。更多信息请关注PHP中文网其他相关文章！
