This dataset was produced in collaboration with the Crop and Soil Science Department of North Carolina State University and the United States Department of Agriculture (USDA). It comprises of 1892 rgb images of plots of soybean fields. The images represent soybean plants having 5 different levels of wilting, each image being assigned a value between 0 and 4 by expert annotators. 0 represents leaves with least wilting while 4 represents the most wilted leaves. The full_data.zip file consists of all images in the dataset. The annotation file dataAnns_goodFiles.csv has all image IDs and their corresponding annotations ranging from 0-4. The last column Annotation has all labels stored from 0-4 for the corresponding image. The treatment_camera column in the csv file refers to the plot number associated with each image. This number is included in the full image id as well. The first part of the number denotes the camera ID while the second part identifies a particular plot ID. For example, 3-106 indicates camera 3, plot 106. The remaining numbers in the image ID denote the date and time when the particular image was captured. For example image 3-106_12_08_2019_06_30_40.jpg denotes camera 3, plot 106 and was taken on 12/08/2019 at 06 hours, 30 mins and 40 seconds.

This work was supported by the United States Department of Agriculture National Institute of Food and Agriculture under the Water CAP program 2018-68011-28372