The yolo format is just images/labels as the 2 top folders, with test/train/val folders in each of these

Each label is a .txt file with
class x_center y_center width height

eg 
0 0.49925 0.8835 0.0225 0.033

This is a single class dataset (0 = cow)

Image patches have had black regions manually removed and have been downsampled to the correct GSD (3 separate operations for PA, PB/C, and all others). 

