Universität Bonn

IGG | Geodesy

Datasets

Template_Pheno4D.jpg
© IGG Geodesy

Pheno4D: A spatio-temporal dataset of maize and tomato plant point clouds for phenotyping and advanced plant analysis

https://www.ipb.uni-bonn.de/data/pheno4d

The dataset contains 7 maize plants measured on 12 days. This gives 84 maize point clouds (about 90 Mio. points). From these, 49 point clouds (about 60 Mio. points) are labeled. Furthermore, the dataset contains 7 tomato plants measured on 20 days (about 350 Mio. points). This gives 140 point clouds from which 77 point clouds (200 Mio. points) are labeled. Note that we provide temporally consistent labels for each point in the clouds. We provide labeled and unlabeled point clouds, the file name indicates whether the point cloud is annotated or not. For example, M01_0313_a.xyz is labeled, M01_0314.xyz is not labeled. For the tomato plant point clouds, each annotated file contains the x,y,z coordinates, and the label associated with the point. For the maize point clouds. Each file annotated contains the x,y,z coordinates, and the 2 labels associated with the point. For both species, if no labels are provided, the files contain only the x,y,z coordinates.

Paper:

  • D. Schunck, F. Magistri, R. A. Rosu, A. Cornelißen, N. Chebrolu, S. Paulus, J. Léon, S. Behnke, C. Stachniss, H. Kuhlmann, and L. Klingbeil, “Pheno4D: A spatio-temporal dataset of maize and tomato plant point clouds for phenotyping and advanced plant analysis,” PLOS ONE, vol. 16, iss. 8, pp. 1-18, 2021. doi:10.1371/journal.pone.0256340
Template_VineSegmentation.jpg
© IGG Geodesy

Segmentation of wine berries

https://doi.org/10.5073/20210308-154150

Dataset contains high resolution images collected with a moving field phenotyping platform, the Phenoliner. The collected images show 3 different varieties (Riesling, Felicia, Regent) in 2 different training systems (VSP=vertical shoot positioning and SMPH= semi minimal pruned hedges), collected in 2 points in time (before and after thinning) in 2018. For each image we provide a manual masks which allow the identification of single berries. The folder contains: 1. List with image details (imagename, acquisition date, year, variety, training system and variety number)and 2. Dataset folder with 2 subfolders, namely 1. img – 42 original RGB images and 2. lbl – 42 corresponding labels (manual annotation, with berry, edge, background definition) The data were used to train a neural network with the main goal to detect single berries in images. The method is described in detail in the specified papers.

Paper:

  • Zabawa, L., Kicherer, A., Klingbeil, L., Töpfer, R., Kuhlmann, H., Roscher, R., (2020) Counting of grapevine berries in images via semantic segmentation using convolutional neural networks. ISPRS Journal of Photogrammetry and Remote Sensing 164, 73–83. https://doi.org/10.1016/j.isprsjprs.2020.04.002
  • Zabawa L, Kicherer A, Klingbeil L, Milioto A, Toepfer R, Kuhlmann H, Roscher R (2019) Detection of single grapevine berries in images using fully convolutional neural networks, The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshop CVPPP, 16-20 June 2019