PlanktonFlow76 - A curated FlowCam dataset for plankton classification

This dataset contains a curated collection of FlowCAM images acquired at INRAE UMR DECOD (Rennes, France) during a multi-year mesocosm experiment in controlled aquatic systems. FlowCAM imaging was used to capture planktonic organisms and particles for ecological monitoring and analysis, and images were segmented using ZooProcess.

The dataset is provided in two forms:

• Raw dataset: ~130,000 images directly collected with the FlowCAM and segmented by ZooProcess (PlanktonFlow-ready starting at the preprocessing step).

• Processed dataset: ~190,000 images obtained after PlanktonFlow pre-processing (augmentation, splitting, and other steps) and ready for downstream analysis. 

Images are organized by taxonomic classes (species) as identified by experts at UMR DECOD. As with any manual classification, occasional errors or ambiguous categories may be present.

This resource can support ecological studies, machine learning model development, and reproducibility of the associated research. For further details, please refer to the associated publication and the open-source codebase available on GitHub.