A temporally transferrable approach for quantifying crop residue cover: linking deep learning of UAV images with satellite-based spectral modelling

This dataset contains the data and code necessary to enable reproducibility of the study “A temporally transferrable approach for quantifying crop residue cover: linking deep learning of UAV images with satellite-based spectral modelling.” It includes UAV RGB imagery and labels used for training the semantic segmentation model for crop residue extraction, together with the corresponding prediction outputs generated by the improved U-Net architecture. In addition, the dataset provides field sampling metadata, including UAV acquisition dates, geographic coordinates, and associated ground-measured crop residue cover (CRC) values. The repository also contains the source code for the improved U-Net model employed for residue segmentation, as well as the scripts used for partial least squares regression (PLSR) modelling and performance evaluation.