Datasets:

Modalities:
Image
Formats:
parquet
Libraries:
Datasets
pandas
License:

You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this dataset content.

Dataset ISIISNet: plankton images captured with the ISIIS (In-situ Ichthyoplankton Imaging System)

Plankton was imaged with an In Situ Ichthyoplankton Imaging System, between surface and ~100m, over 10 days in July 2016, in the North Western Mediterranean Sea. This deployment was the core of the VISUFRONT cruise. The image generated by the linescan, shadowgraph camera of ISIIS were processed with the custom software apeep and regions of interest, targeted to be planktonic organisms by a deep segmenter, were extracted. The 408,166 resulting objects were sorted by a limited number of operators, following a common taxonomic guide, into 32 taxa, using the web application EcoTaxa http://ecotaxa.obs-vlfr.fr. For the purpose of training machine learning classifiers, the images in each class were split into training, validation, and test sets, with proportions 70%, 15% and 15%.

Details

  • train split means (RGB): [0.9464540308383366]
  • train split standard deviations (RGB): [0.16473865521613654]

Samples per class for split train

0: Acantharea                2110.00
1: Actinopterygii            276.00
2: Annelida                  130.00
3: Appendicularia            419.00
4: Appendicularia_body       3265.00
5: Appendicularia_like_body  835.00
6: Aulacanthidae             1952.00
7: Bacillariophyceae         1859.00
8: Chaetognatha              240.00
9: Cnidaria                  130.00
10: Copepoda                ▇▇ 12073.00
11: Crustacea                981.00
12: Ctenophora               1362.00
13: Doliolida               ▇ 7079.00
14: Eumalacostraca           95.00
15: Harpacticoida            646.00
16: Mollusca                 186.00
17: Pyrocystis               2567.00
18: Rhizaria                 2064.00
19: Rhopalonematidae         423.00
20: Siphonophorae            1699.00
21: colonial_colodaria       290.00
22: detritus                ▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 321335.00
23: ephyra                   480.00
24: house                    92.00
25: like_Acantharea          89.00
26: like_Copepoda           ▇▇▇ 19196.00
27: other_living             987.00
28: part_Cnidaria            1281.00
29: solitaryblack            454.00
30: streak                  ▇▇▇ 23501.00
31: vertical line            70.00

Reference

Panaïotis, T., Caray-Counil, L., Jalabert, L., & Irisson, J.-O. (2024). ISIISNet: plankton images captured with the ISIIS (In-situ Ichthyoplankton Imaging System). SEANOE. https://doi.org/10.17882/101950

BibTEX

@article{dataset:isiisnet,
  title       = {ISIISNet: plankton images captured with the ISIIS (In-situ Ichthyoplankton Imaging System)},
  author      = {Panaïotis, Thelma and Caray-Counil, Louis
                and Jalabert, Laetitia and Irisson, Jean-Olivier},
  year        = 2024,
  journal     = {SEANOE},
  doi         = {10.17882/101950},
  affiliation = {Sorbonne Université, CNRS, Laboratoire d'Océanographie de Villefanche, LOV, 06230 Villefranche-sur-mer, France.
    Sorbonne Université, CNRS, Institut de la Mer de Villefranche, IMEV, 06230 Villefranche-sur-Mer, France.}
}

Usage

from datasets import load_dataset

dataset = load_dataset("project-oceania/isiisnet")
Downloads last month
1

Collection including project-oceania/isiisnet