Dataset of 100 simulated root images, with different resolutions

All images were analysed with WinRhizo^TM 2013, Regent method (Régent Instruments Inc 2013) and ImageJ (1.51j8, Schneider, Rasband, & Eliceiri, 2012) with the macro IJ_Rhizo (IJ_Rhizo_v0beta). We conducted batch analyses with the grey level thresholds set to automatic at all four resolutions (200, 400, 800 and 1200 DPI). We further analysed the images at 800 dpi and 1200 dpi with the threshold value manually set to 144. We considered this value suitable after visual inspection of 10 images at different thresholds using the threshold slider of WinRhizo^TM. Since no debris was simulated, we used no automatic debris correction. For WinRhizo^TM, the boundaries of diameter classes were set from 0.1 to 1.9 mm (with an increment of 0.1 mm).

The dataset contains:

- the root images in jpg format (images.zip)

- the root architecture data in rsml format (rsml.zip)

- the ground-truth value for each image (all_root_data_2D.csv)

- the ground-truth value for each root in each image (ground_truth_data_2D.csv)

- the estimation given by WinRhizo and IJ_Rhizo (all_analyses_2D.txt and all_analyses_fixed.txt)

- the different codes used for the analysis (*.R)

Dataset used in : Accuracy of image analysis tools for functional root traits: A comment on Delory et al., 2017 from Rose and Lobet (2019), Methods in Ecology and Evolution. 

The images in this dataset are a subset of the dataset here: https://doi.org/10.5281/zenodo.208214