PlanktoShare: A large (50k+) and FAIR learning set for the Plankton Imager (Pi-10) for the Greater North Sea and NE Atlantic, based on a new flexible classification protocol
Authors/Creators
-
van Walraven, Lodewijk
(Data collector)1
- Scott, James (Data collector)2
- Pitois, Sophie (Data collector)2
- Ribeiro, Joseph (Related person)2
- Close, Hayden (Data collector)2
- Liszka, Cecilia (Related person)3
- Fileman, Elaine (Related person)4
- Hoekendijk, Jeroen (Related person)5
- Hovenkamp, Pieter (Related person)6
- Jak, Robbert (Data collector)5
- van Dalen, Joost (Related person)5
- van Oevelen, Dick (Researcher)6
- 1. Rijkswaterstaat
-
2.
Centre for Environment, Fisheries and Aquaculture Science, Lowestoft Laboratory
-
3.
British Antarctic Survey
-
4.
Plymouth Marine Laboratory
-
5.
Wageningen Marine Research (Den Helder)
-
6.
Royal Netherlands Institute for Sea Research
Description
This is the accompanying dataset to the submitted paper "PlanktoShare: A large (50k+) and FAIR learning set for the Plankton Imager (Pi-10) for the Greater North Sea and NE Atlantic, based on a new flexible classification protocol." by Lodewijk Van Walraven, James Scott, Sophie Pitois, Joseph Ribeiro, Hayden Close, Cecilia M. Liszka, Elaine Fileman, Jeroen Hoekendijk, Pieter Hovenkamp, Robbert Jak, Joost van Dalen, Dick van Oevelen.
The dataset contains labeled images of plankton, detritus and other particles collected using the Pi-10 Plankton Imager (Plankton Analytics) in the North East Atlantic area on the research vessels Tridens II and CEFAS Endeavour. The labeling convention is explained in the paper and in the file PlanktoShare_readme.txt.
Code for training and inference on new unseen images is publicly available at: github.com/geoJoost/plankton_imager_classifier, to allow iterative model development by other researchers
Files
PlanktoShare.zip
Additional details
Software
- Repository URL
- https://github.com/geoJoost/planktoshare
- Programming language
- Python
- Development Status
- Active