2018 Data Science Bowl nuclei segmentation (BBBC038v1) training set as COCO polygons
- Provider
- Broad Institute (Caicedo et al., BBBC038); COCO conversion by DataTorch
- License
- CC0-1.0 (source)
- Size
- 673 files
- Format
- JSON, PNG (in sampled files)
- Stored at
- Hugging Face (https://huggingface.co/datasets/DataTorchio/nuclei-segmentation-dsb2018)
- Access
- Free and public. No account with the data provider is needed.
- Category
- Life Sciences
- Source
- https://doi.org/10.1038/s41592-019-0612-7
About this dataset
The stage 1 training set of the 2018 Data Science Bowl, image set BBBC038v1 from the Broad Bioimage Benchmark Collection: 670 microscopy images with 29,430 individually segmented nuclei. Images span fluorescence, brightfield and histology stains at several magnifications, from human, mouse and fly samples, and were assembled to train generalist nucleus segmentation models. In this copy the official per-nucleus masks are converted to polygon annotations in one COCO-format annotations.json (12 of 29,442 masks were too small to outline and were dropped), with the images under images/; about 70 MB. Load with pycocotools, torchvision or any COCO-compatible detection and instance segmentation tool, for example to train Mask R-CNN, Cellpose or StarDist style models. The copy is maintained by DataTorch, an annotation platform. The Broad Institute releases BBBC038v1 under CC0 (contributors have waived copyright); cite Caicedo et al., "Nucleus segmentation across imaging experiments: the 2018 Data Science Bowl", Nature Methods 16, 1247-1253 (2019).
How to load this dataset in Python
from huggingface_hub import hf_hub_download
path = hf_hub_download(repo_id="DataTorchio/nuclei-segmentation-dsb2018", filename="images/00071198d059ba7f5914a526d124d28e6d010c92466da21d4a04cd5413362552.png", repo_type="dataset")
print(path)Data structure
File types in sampled files: JSON, PNG
Files
Example file: images/00071198d059ba7f5914a526d124d28e6d010c92466da21d4a04cd5413362552.png
Analyze this dataset without downloading it
On Scigantic, 2018 Data Science Bowl nuclei segmentation (BBBC038v1) training set as COCO polygons opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.
This page is also available as Markdown and as JSON.
Related Life Sciences datasets
- Ocean Biodiversity Information System (OBIS) species occurrence data
- ASEM Annotated FIB-SEM Cell Volumes
- Heliconius Mimicry Gold Standard: Dorsal Butterfly Images with Bird and Butterfly Visual Acuity Transforms
- DHARANI Developing Human-Brain Atlas
- ZINC Database
- AlphaFold Protein Structure Database
- NIS3D: densely annotated 3D nuclei segmentation benchmark in zebrafish, Drosophila and mouse embryos
- kabr-tools Methodology Dataset: Focal, Scan and Drone Behavior Observations of Grevy's Zebras, Plains Zebras and Giraffes