TimeSen2Crop
TimeSen2Crop is a pixel based dataset made up of more than 1 million samples of Sentinel 2 Time Series associated to 16 crop types, during an agronomic year ranging from September 2017 to August 2018. The dataset contains atmospherically corrected samples, as well as information related to snow, clouds and shadows.
This benchmark dataset has been developed in the framework of the ExtremeEarth project, which received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 825258.
The dataset is organized hierarchically. The data are organized per Sentinel 2 tiles, in 16 folders (15 tiles plus one tile for the 2019 agronomic year). Each Sentinel 2 folder
contains 16 sub-folders where samples for each crop type are stored. The Sentinel 2 folder contains also a csv file with the acquisition dates for the samples in that particular tile, ordered from the oldest acquisition to the newest. Inside the crop type sub-folders, each labeled sample is stored in a csv file containing the multispectral temporal signature. The csv files provides a matrix where each row indicates the acquisition date and each column indicates the spectral band associated, namely blue (B2), green (B3), red (B4), the four vegetation red edge (B5, B6, B7 and B8A) and the two short wave infrared (SWIR) (B11 and B12). The last column contains information regarding the condition of the pixel: clear (0), cloud (1), shadow (2) or snow (3).
