Annotated IFCB plankton images from the Mediterranean Sea

Area of Science:

• Marine Biology
• Ecology
• Data Science

Background:

• The Imaging FlowCytobot (IFCB) automates plankton monitoring, enhancing research resolution.
• Training machine learning algorithms for IFCB analysis requires extensive, labeled image datasets.
• Existing datasets are limited, particularly lacking Mediterranean coverage.

Purpose of the Study:

• Introduce MedPlanktonSet, a novel dataset of taxonomically annotated IFCB images.
• Address the scarcity of shared datasets for IFCB machine learning model development.
• Provide a valuable resource for plankton research in the Western Mediterranean Sea.

Main Methods:

• Collected 77,271 IFCB images from November 2022 to February 2025.
• Established six sampling stations in the Gulf of Naples, Western Mediterranean Sea.
• Classified images into 139 distinct plankton categories.

Main Results:

• Successfully curated MedPlanktonSet, a comprehensive dataset of annotated IFCB images.
• The dataset includes associated features for each image.
• Achieved taxonomic annotation for 139 categories.

Conclusions:

• MedPlanktonSet will accelerate the development of machine learning classifiers for plankton identification.
• The dataset supports 3D plankton reconstructions and trait-based ecological studies.
• Facilitates advancements in understanding plankton biodiversity and ecology through enhanced IFCB utilization.