IFCB phytoplankton anomaly dataset (IFCB-PAD)

IFCB phytoplankton anomaly dataset (IFCB-PAD) contains over 6200 manually annotated and expert-validated phytoplankton images (9 plankton classes) with anomalies such as parasites. The dataset with bounding box annotations is available in both COCO and YOLO format. OK images (no anomalies) are derived from SYKE-plankton_IFCB_2022 dataset (https://doi.org/10.23728/b2share.abf913e5a6ad47e6baa273ae0ed6617a) and NOK images consist of unpublished data measured in the 2021 on Utö station using Imaging FlowCytobot (IFCB, McLane Research Laboratories, Inc., U.S., Olson and Sosik, 2007). The plankton class list: - Aphanizomenon - Centrales - Dolichospermum - Chaetocero - Nodularia - Pauliella - Peridiniella Chain - Peridiniella Single - Skeletonem If you use this dataset in your research, we kindly ask that you reference the following paper: Bilik, S., Baktrakhanov, D., Eerola, T., Haraguchi, L., Kraft, K., Wyngaert, S.V.D., Kangas, J., Sjöqvist, C., Madsen, K., Lensu, L. and Kälviäinen, H., 2023. Towards Phytoplankton Parasite Detection Using Autoencoders. arXiv preprint arXiv:2303.08744.