Learning county from pixels: corn yield prediction with attention-weighted multiple instance learning

Remote sensing has become a promising tool in yield prediction. Most prior work using satellite imagery for county-level corn yield prediction spatially aggregates all pixels within a county into a single value, potentially overlooking the detailed information and valuable insights offered by more granular data. This research studies each county at the pixel level and applies multiple instance learning to leverage the detailed information within each county. In addition, our method addresses the "mixed pixel'' problem caused by the inconsistent resolution between feature datasets and crop masks, which may introduce noise into the model and make corn yield prediction more difficult. Specifically, the attention mechanism is utilized to automatically assign weights to different pixels, which can mitigate the influence of the "mixed pixel'' problem. The experimental results show that the developed model outperforms four other machine learning models over the past 5 years in the U.S. Corn Belt and demonstrates its best performance in 2022, achieving a coefficient of determination (${R<^>2}$R2) value of 0.84 and a root mean square error ($RMSE$RMSE) value of 0.83. This paper demonstrates the advantages of our approach from both spatial and temporal perspectives. Furthermore, through an in-depth study of the relationship between mixed pixels and attention, it is verified that our approach can capture important feature information while filtering out noise from mixed pixels.