AgroBench：弱监督作物产量学习的可复现多模态基准
原标题：AgroBench: A Reproducible Multimodal Benchmark for Weakly Supervised Crop Yield Learning from County Statistics and Pixel Observations
AI 摘要
研究团队发布 AgroBench，一个可复现的多模态基准，用于弱监督作物产量学习。该基准将美国县级 USDA 作物产量统计与 Sentinel-2 光学影像、Sentinel-1 SAR、气候和地形数据结合，生成像素级作物时序，每个像素序列以县级产量作为弱监督信号。数据集包含 788,654 个作物像素、超 1300 万条观测，覆盖 2017–2024 八个生长季、五种主要作物和 5,107 个县-年组合，并提供留一年评估协议与基线结果。
正文节选
AgroBench: A Reproducible Multimodal Benchmark for Weakly Supervised Crop Yield Learning from County Statistics and Pixel Observations Abstract Reliable agricultural yield statistics are typically reported at coarse administrative scales, whereas modern geospatial machine learning methods require spatially explicit, pixel-level supervision. This mismatch has limited the development of large-scale benchmarks for crop yield learning using multimodal Earth observation data. A reproducible benchmark