Contains the codes from the paper "Self-Supervised Pretraining for Fine-Grained Plankton Recognition".
This repository contains code for pretraining a vision transformer using masked autoencoder (MAE) utilizing multiple plankton datasets.
- K. He, et al., "Masked autoencoders are scalable vision learners," CVPR 2022
In the paper, multiple public plankton datasets were utilized for the pretraining. The full list with the references is shown below: Table: Summary of the datasets used for pretraining.
| Dataset | Plankton Type | # of Species | # of Images | Link |
|---|---|---|---|---|
| Kaggle-Plankton (Cowen2015) | zooplankton | 121 | 130,000 | Link |
| Lake Zooplankton (kyathanahally2021deep) | zooplankton | 35 | 18,000 | Link |
| SYKE-Plankton-ZooScan_2024 (zooscan2024) | zooplankton | 20 | 24,000 | Link |
| PMID2019 (li2020developing) | phytoplankton | 24 | 14,000 | Link |
| SYKE-Plankton-IFCB_2022 (syke2022) | phytoplankton | 50 | 63,000 | Link |
| UDE Diatoms in the Wild 2024 (Kloster2024) | phytoplankton | 611 | 84,000 | Link |
| DAPlankton (batrakhanov2024daplankton) | phytoplankton | 44 | 112,000 | Link |
| Total | 443,000 |
@misc{opensetplankton2025,
author={Joona Kareinen and Tuomas Eerola
and Kaisa Kraft and Lasse Lensu
and Sanna Suikkanen and Heikki K\"{a}lvi\"{a}inen},
title={Self-Supervised Pretraining for Fine-Grained Plankton Recognition},
year="2025"
}