Datasetpublic

patcdaniel/synchro-April2025-cluster-labeled-highMag

IFCB Plankton Labeled (Cluster-Sorted) This dataset contains labeled images of phytoplankton collected with the Planktivore Imaging System. Images were preprocessed with a zero-padding and resized to the standard size used for ViT_b_16 The dataset was originally constructed by clustering unlabeled ROI images using deep features from a ViT model.Clusters were then saved locally and manually curated into taxonomic labels and higher-order groups. Dataset Summary… See the full description on the dataset page: https://huggingface.co/datasets/patcdaniel/synchro-April2025-cluster-labeled-highMag.

sourceHugging Facecc-by-4.0updated 1y agoView on Hugging Face

0likes15downloads

Dataset Card

IFCB Plankton Labeled (Cluster-Sorted)

This dataset contains labeled images of phytoplankton collected with the Planktivore Imaging System. Images were preprocessed with a zero-padding and resized to the standard size used for ViT_b_16

The dataset was originally constructed by clustering unlabeled ROI images using deep features from a ViT model. Clusters were then saved locally and manually curated into taxonomic labels and higher-order groups.

Dataset Summary

—Modality: Images (PNG)
—Source: Planktivore ROI captures
—Curation process:
—Extracted deep features with a ViT backbone.
—Applied clustering (UMAP + HDBSCAN) to group morphologically similar images.
—Exported clusters to local folders.
—Manually reviewed and sorted each cluster into taxonomic categories (label) and broader groups (group).

Columns

—image: The plankton ROI image.
—label: Fine-grained label (taxon).
—group: Higher-order grouping (e.g. diatoms, dinoflagellates, ciliates).

Example

python

from datasets import load_dataset

ds = load_dataset("patcdaniel/synchro-April2025-cluster-labeled-highMag")
sample = ds["train"][0]
sample["image"].show()
print("Label:", ds["train"].features["label"].int2str(sample["label"]))
print("Group:", ds["train"].features["group"].int2str(sample["group"]))