Dataset of article "Decision support for the identification of testate amoebae in microscopy images to detect peat presence in horticultural substrates"

This data set was used in the following article: "Decision support for the identification of testate amoebae in microscopy images to detect peat presence in horticultural substrates" by Serge Zaugg¹, Camille Vögeli², Lena Märki¹, Clément Duckert² and Edward Mitchell². ¹Swiss Federal Institute of Metrology METAS. ²Laboratory of Soil Biodiversity, University of Neuchâtel.

Data overview

The data is in the zipped folder testate_amoebae_in_peat. yolov5pytorch. This data is in YOLOv5 format which is compatible with YOLOv8 models. Images have sizes 1728 x 1296, 1728 x 1080, or 1728 x 976 pixels. The training set comes from microscopy images of 33 substrate samples and the validation set from 16 independent samples. The latter can be used to assess generalization performance. Data in the validation folder is referred to as test set in the article. The first seven digits of the files contain the sample ID. As an overview, we provide two large PNG images that show the content of the bounding boxes organized by class.

Data structure

testate_amoebae_in_peat.yolov5pytorch

|-- data.yaml   (link between numeric class codes and class names)

|-- train

    |-- images  (5928 images, pre-augmented see section 2.5)

    |-- labels  (5928 text files with labels)                       

|-- valid

    |-- images  (2415 images)

    |-- labels  (2415 text files with labels)

Funding Statement

This study was commissioned by the Federal Office for the Environment (FOEN), CH-3003 Bern. The FOEN is an agency of the Federal Department of the Environment, Transport, Energy and Communications (DETEC). Disclaimer: This study was prepared under contract to the Federal Office for the Environment (FOEN). The contractor bears sole responsibility for the content.