Example datasets for training and testing ag-net for crop recognition
This is related to ESIP Google Summer of Code project idea - AgNet
The input folder contains the 8bit version of seven bands of Landsat 8. Each file name is constructed in this pattern:
lc8_{observation_date}_{row number}_{column number}_{band number}.tif
The input_32bit contains the original values of Landsat 8. The valid value range is 0-10000. The file name structure is the same.
The input_rgb contains a RGB composite of Landsat 8 band 2, 3, 4. The file name structure is:
lc8_{observation_date}_{row number}_{column number}_rgb.tif
The target folder contains the corresponding tiles. The file name is composed of:
cdl_{landsat observation date}_{row number}_{column number}.tif
The value of CDL ranges from 0 to 255. Each value represents a crop type or non-crop land cover. The CDL value explanation table is here.
Please cite this dataset by:
Ziheng Sun. (2019). Ag-Net Sample Training Dataset [data files and tutorial notebook]. https://github.com/ZihengSun/Ag-Net-Dataset.