DiPAS

This dataset contains the training and testing dataset (images and masks) used for grass detection and segmentation. There are 100 images and masks in the training set; and 20 images and masks in the testing set. These data were collected from the Google Earth, generated by zooming in the map and randomly capturing 120 images of a specific region (Oxford City in this study). The labels were generated using an annotation tool provided by a cloud machine-learning platform called Amazon SageMaker by one annotator.