PlanktonSet 1.0: Marine Imagery from the Straits of Florida

Subtropical Straits of Florida imagery data collected from May to June 2014 using the In Situ Ichthyoplankton Imaging System (ISIIS-2). This subset comprises 121 plankton classes, split evenly into training and test sets, and was used for the inaugural National Data Science Bowl. The dataset is intended as a standard benchmark for machine learning and computer vision tasks in plankton image classification.

Use Cases

• Training image classification models on 121 distinct plankton classes.
• Benchmarking computer vision algorithms for fine-grained biological recognition.
• Studying taxonomic or structural relationships in marine plankton based on image features.
• Developing methods for automated plankton identification from in-situ imagery.

Strengths

• Contains 121 distinct plankton classes, providing a multi-class classification challenge.
• Data is pre-split into 50:50 train and test sets, facilitating direct benchmarking.
• Collected using a standardized in-situ imaging system (ISIIS-2) over a defined period in 2014.

Limitations

• Key metadata such as exact row count, file size, and license are missing from all sources.
• The dataset is a subset of a larger collection, so the full scope of available imagery is unknown.
• Last update dates conflict between platforms (2014 vs. 2026), casting doubt on metadata maintenance.

Provenance

Source

National Oceanic and Atmospheric Administration, Department of Commerce

Collection Method

Imagery collected in situ using the ISIIS-2 system as part of an NSF-funded project.

Time Range

2014-05-28 to 2014-06-14

Freshness

2026-03-05 23:59:35.281394

Geography

Subtropical Straits of Florida