mpg-ranch/leafy_spurge

Background

This dataset comprises 1.3 cm resolution aerial images of grasslands in western Montana, USA, captured by a commercial drone. Many scenes contain leafy spurge (Euphorbia esula), introduced to North America, now widespread in rangeland ecosystems, which is highly invasive and damaging to crop production and biodiversity. Technicians surveyed 1000 points in the study area, noting spurge presence or absence, and recorded each point’s position with precision global navigation satellite systems. We extracted crops from an orthoimage at these locations. We publicly release these images and metadata as a Hugging Face Dataset, accessible in one line of code. Our aim is to invite the research community to develop classifiers as early warning systems for spurge invasion. We established baselines for state-of-the-art vision models, achieving 0.85 test accuracy, demonstrating the feasibility yet difficulty of this classification task.

Please visit our website for updates and related work: Leafy Spurge Dataset

Data Access

Please install the following dependencies to load Leafy Spurge Dataset:

pip install datasets pillow

Download the dataset as follows:

#datasets module documentation at:
#https://huggingface.co/docs/datasets/en/quickstart
from datasets import load_dataset

#39x39 pixel train and test data
crop_train = load_dataset('mpg-ranch/leafy_spurge', 'crop', split='train') 
crop_test = load_dataset('mpg-ranch/leafy_spurge', 'crop', split='test') 

#1024x1024 pixel train and test data
context_train = load_dataset('mpg-ranch/leafy_spurge', 'context', split='train') 
context_test = load_dataset('mpg-ranch/leafy_spurge', 'context', split='test') 

#unlabeled orthomosaic
orthomosaic = load_dataset('mpg-ranch/leafy_spurge', 'unlabeled', split='train') 

# Get the first observation from the crop training set
first_observation = crop_train[0]

# Extract the image and metadata
image = first_observation['image']
metadata = {key: value for key, value in first_observation.items() if key != 'image'}

# Print the metadata and show image
print("Metadata for the first observation:")
for key, value in metadata.items():
    print(f"{key}: {value}")

image.show()

Suggested Citation

To cite this Hugging Face Dataset repository, we suggest:

@misc{doherty2025spurgedata,
  title={Leafy Spurge Dataset},
  author={
    Doherty, Kyle and 
    Gurinas, Max and 
    Samsoe, Erik and 
    Casper, Charles and 
    Larkin, Beau and 
    Ramsey, Philip and 
    Trabucco, Brandon and 
    Salakhutdinov, Ruslan
  },
  year={2025},
  howpublished = {Hugging Face Datasets, \url{https://huggingface.co/datasets/kylewdoherty/leafy-spurge}},
  doi={10.57967/hf/2508},
  }

This dataset accompanies our manuscript entitled “Ground-truthed and high-resolution drone images of the leafy spurge weed plant (Euphorbia esula)”, which has been provisionally accepted (pending minor revisions) at Scientific Data. Upon final publication, we will update this citation to reflect the journal-assigned DOI.

@article{doherty2025spurgedescriptor,
  title = {Ground-truthed and high-resolution drone images of the leafy spurge weed plant (Euphorbia esula)},
  author = {Doherty, Kyle and 
    Gurinas, Max and 
    Samsoe, Erik and 
    Casper, Charles and 
    Larkin, Beau and 
    Ramsey, Philip and 
    Trabucco, Brandon and 
    Salakhutdinov, Ruslan},
  journal = {Scientific Data},
  year = {2025},
  doi = {UPDATES PENDING}
}