Training image data for: Environmental and ecological drivers of harmful algal blooms in the Southern California Bight

The datasets include example images of marine protists and copepods that were used for training of machine classifiers in Kenitz et al. (in review). The images were collected by the Scripps Plankton Camera (SPC; http://spc.ucsd.edu/), located at the end of the Scripps Pier, La Jolla, CA, at the tide-dependent depth of approximately 3-4 m. The Scripps Plankton Camera system includes two free-space, darkfield microscopes: (i) micro-SPC equipped with a 5x objective for imaging of marine protists, and (ii) mini-SPC equipped with a 0.5x objective for imaging larger zooplankton. The images in this database were used for training of two distinct classifiers: 1) Multi-class classifier for classification of marine protists: Images for training of the multiclass classifier include examples of marine protists and other particles detected by the camera with the major axis in the range of 30-1000 microns. These images were collected by the micro-SPC. 2) Binary classifier for identification of marine copepods: Images for training of the binary classifier include example images of marine copepods and non-copepods with the major axis in the range of 1-5 mm. These images were collected by the mini-SPC. Example images for the non-copepod class include everything not specifically identified as a copepod, such as cladocerans, euphausiids, appendicularians, large diatom chains and other unidentifiable particles.