Work in progress (last modified 2022-03-31). Manually sorted and annotated mesozooplankton images obtained using a FlowCam Macro Yokogawa Fluid Imaging Technologies (camera and microscope system) equipped with a 0.5x objective lens and a 5000µm deep field-of-view flowcell (magnification = 12.5). The size of one pixel is ~9.05 µm. The images have been sorted into a taxonomical folder structure according to phylogeny (e.g. Family – Order – Genus - Species). This version (last modified 2022-03-31) is incomplete in the phylogenetic organization of the folders. Plankton sizes range from 150-3000µm ABD (Area Based Diameter).
The plankton samples were WP2 plankton net hauls (180 µm mesh). Samples were fixed in 4% formaldehyde, and subsequentially diluted in freshwater during imaging. Samples originated mostly from the North-Atlantic Ocean, i.e. Norwegian Sea and North Sea. A few samples originated from Antarctica.
This version of the dataset needs further improvement (number of images in each class) in order to train a machine learning model and apply it for the classification of real plankton samples. As a preliminary data-set it was intended to investigate machine-learning algorithms. There may be misplaced images and misannotations.