The Plantorgan hunter dataset

These organelles exist in various sizes and have different characteristics at different nanometer levels. All images have a resolution of 3730 × 4800, 686 dpi, and a storage size of 17 MB. The original image format is tif format, saved in the image folder.The ".json" data format manually annotated by LabelMe software is stored in the jsons compressed package.

These json data can be opened and re-changed through LabelMe.The label image format is png format, including the masks of organelles, saved in the label folder. Four plant cell organelles (chloroplast, mitochondria, nucleus, and vacuole) were labeled.The splits folder holds the dataset partitions for training OrgSegNet.3D-Cell Stacks.zip archive holds images of sequence slices from multiple plant cells, which can be used to reconstruct 3D organelles.The Krippendorff s_Alpha.zip archive contains a sub-dataset for calculating Krippendorff's Alpha, annotated by 4 botany majors.The Unseen data folder contains the external data (Unseen data) used in the PlantOrganelle paper (rice mutants, M. sativa and E. haichowensis under different treatement).This dataset will be continuously updated in the future to promote a better understanding of plant subcellular biology.If you use this dataset, please cite: Feng, X., Yu, Z., Fang, H. et al. Plantorganelle Hunter is an effective deep-learning-based method for plant organelle phenotyping in electron microscopy. Nat.