imageomics/NEON-plant-subplot-pilot

Dataset Card for NEON Plant Presence and Percent Cover Subplot Pilot Images

Pilot dataset for using subplot images from the National Ecological Observatory Network (NEON) to detect plant diversity.

Dataset Details

Dataset Description

• Repository: PlotDiversiVision

This dataset contains plant diversity images from two NEON sites: CPER and SCBI. Each image captures a 1-square-meter subplot labeled with plant species. NEON curated subplot images to estimate plant diversity. This dataset serves as a pilot study for standardizing the curation pipeline and performing empirical species classification experiments using BioCLIP 2. The code used for processing and producing predictions is available at PlotDiversiVision.

Supported Tasks and Leaderboards

This dataset has multiple plant species labels associated with each image. It supports the task of multi-label image classification.

Dataset Structure

The dataset includes images from 7 NEON plots and their labels aligned with BioCLIP 2.

images/
    metadata.csv
    <plot_1>/
        <plot_1>_PlantDiversity_<subplot_1>_<time>-straightened.jpg
        <plot_1>_PlantDiversity_<subplot_2>_<time>-straightened.jpg
        ...
        <plot_1>_PlantDiversity_<subplot_n>_<time>-straightened.jpg
    <plot_2>/
        <plot_2>_PlantDiversity_<subplot_1>_<time>-straightened.jpg
        <plot_2>_PlantDiversity_<subplot_2>_<time>-straightened.jpg
        ...
        <plot_2>_PlantDiversity_<subplot_n>_<time>-straightened.jpg
    ...
    <plot_7>/
        <plot_7>_PlantDiversity_<subplot_1>_<time>-straightened.jpg
        <plot_7>_PlantDiversity_<subplot_2>_<time>-straightened.jpg
        ...
        <plot_7>_PlantDiversity_<subplot_n>_<time>-straightened.jpg
labels/
    <plot_1>_subplot_labels.csv
    <plot_2>_subplot_labels.csv
    ...
    <plot_7>_subplot_labels.csv

Data Instances

metadata.csv links the image files to their plot_id and subplot_id.

All images are named with plot_id and subplot_id. The labels are included as a list of species for each subplot. Subplot IDs are in the format of X_Y_Z, where X and Z are used for indexing, and Y is the subplot size. The subplot labels with Y=1 are corresponded with images.

Data Fields

metadata.csv:

• file_name: The path to the image file.
• plot_id: The ID for the plot, e.g., CPER_001.
• subplot_id: The ID for the subplot in the plot, e.g., 31_1_1.

plot_N_subplot_labels.csv:

• plotID: The ID for the plot, e.g., CPER_001.
• subplotID: The ID for the subplot in the plot, e.g., 31_1_1.
• original_labels: The original species label list provided by NEON technicians.
• resolved_labels: Resolved binomial name list produced by TaxonoPy.
• resolved_scientific_names: Resolved scientific name list produced by TaxonoPy.
• resolved_taxonomic_labels: Resolved taxonomic label list produced by TaxonoPy.
• taxonopy_resolution_status: The resolution status for every label in the list.
• label_count: Number of species labeled by NEON technicians for every image.
• unmapped_original_labels: The original labels that cannot be mapped by the species lists.

Data Splits

This dataset is used fully for testing.

Dataset Creation

Curation Rationale

This dataset is a pilot study in using NEON's subplot images. The 2 sites (CPER and SCBI) are selected based on their clean square boundary and diversity in plant distribution. We randomly select 3 and 4 plots from these 2 sites, respectively.

Source Data

Data Collection and Processing

The images in this dataset are pre-processed from original NEON images using imageonline to create a perspective crop.

Who are the source data producers?

The source data is provided by NEON.

Annotations

We first used imageonline to create image crops. Then we processed the original labels with TaxonoPy to retrieve the 7-rank taxonomic label and make sure the labels are aligned with BioCLIP 2 training.

Annotation process

We curated species list from four different sources:

• Observed species reported by NEON
• State-level species list from CONUS
• Regional species list from BONAP
• Regional species list from GBIF The species list is processed through TaxonoPy. The derived species list is deduplicated. Entries with a less-than-5-level taxonomic label are removed.

Who are the annotators?

Sooyoung Jeon, Braedon Lineman, and Arpita Chowdhury used imageonline to create crops. Jianyang Gu used TaxonoPy to create the aligned species labels.

Bias, Risks, and Limitations

This dataset is an initial curation from a small fraction of NEON subplot images. It serves as a proof-of-concept to build up processing pipelines and use BioCLIP 2 to predict plant species. Further work will release a larger curation of subplot images that are ready for larger-scale benchmarking and method development.

The labels are processed with TaxonoPy, which mainly aims to align with BioCLIP 2. Therefore, there are risks that the labels are not aligned with specific uses.

Licensing Information

Images and graphics may have been generously contributed to NEON, or generated by NEON, to promote education and research. Unless stated otherwise, images are made available under the Creative Commons Attribution 2.0 (CC BY 2.0). Users are allowed to copy, transmit, reuse, and/or adapt content, as long as attribution regarding the source of the content is made. If the content is altered, transformed, or enhanced, it may be re-distributed only under the same or similar license by which it was acquired.

Citation

BibTeX

@misc{neonPlantSubplotPilot,
  author = {Dave Barnett and Arpita Chowdhury and Jianyang Gu and Leanna House and Sooyoung Jeon and Eugene Law and Braedon Lineman},
  title = {NEON Plant Presence and Percent Cover Subplot Pilot Images},
  year = {2026},
  url = {https://huggingface.co/datasets/imageomics/neon-plant-subplot-pilot},
  doi = {<doi once generated>},
  publisher = {Hugging Face}
}

Acknowledgements

This work was supported by the Imageomics Institute, which is funded by the US National Science Foundation's Harnessing the Data Revolution (HDR) program under Award #2118240 (Imageomics: A New Frontier of Biological Information Powered by Knowledge-Guided Machine Learning). Any opinions, findings and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

This material is based in part upon work supported by the National Ecological Observatory Network (NEON), a program sponsored by the U.S. National Science Foundation (NSF) and operated under cooperative agreement by Battelle.

Dataset Card Authors

Jianyang Gu

Dataset Card Contact

gu.1220@osu.edu