An iterative energy-based multimodal transformer for joint retrieval of wheat soil moisture, leaf area index, and plant height from Sentinel-1 and Sentinel-2 time series
基于迭代能量多模态Transformer的Sentinel-1与Sentinel-2时序联合反演小麦土壤水分、叶面积指数和株高
机构 * Department of Urban and Environmental Policy and Program, Tufts University(Tufts大学城市与环境政策与项目系) ; Electrical and Computer Engineering Department, Mississippi State University(密苏里州立大学电气与计算机工程系) ; Department of Physics, Indian Institute of Technology (BHU)(印度理工学院(BHU)物理系) ; Institute of Environment and Sustainable Development, Banaras Hindu University(巴纳尔斯赫尔大学环境与可持续发展研究所)
专题命中 跨模态检索 :multimodal(title,abstract);multi-modal(abstract);分类 cs.CV
AI总结 提出迭代能量Transformer(iEBT),通过嵌入多模态预测器并迭代更新目标向量,从Sentinel-1和Sentinel-2时序数据中联合反演土壤水分、叶面积指数和株高,在印度瓦拉纳西实测数据上取得高精度,并利用终端能量作为质量诊断指标。