Hugging Face's logo

Datasets:
LynnMass
/
tilebac-flag-dataset

Tasks:
Image Segmentation
Modalities:
Image
Text
Formats:
imagefolder
Languages:
English
Size:
1K - 10K
Tags:
flagella
bacteria
Libraries:
Datasets
License:
Dataset card Data Studio Files
xet
 Community
1
Dataset Viewer
Auto-converted to Parquet Duplicate
Search is not available for this dataset
image
imagewidth (px)
1.02k
4.1k
End of preview. Expand in Data Studio

Background

Overview This dataset contains ultralow-dose cryoEM montage tile images of the bacteria Pantoea sp. YR343. Segmentation models from YOLOv11, YOLO26, U-Net, Detectron2 and SAM3 have been fine-tuned to predict bacterial inner membranes and outer membranes. This bacterial membrane dataset is a benchmark dataset to challenge current AI workflows in rapid, seamless montage stitching and stitched segmentation in extremely noisy ultralow-dose cryoEM images. Bacterial flagella low-dose cryoEM images have also been added as a dataset challenge to segmenting high-boundary thin objects in noisy low-dose cryoEM images.

Reference

Please cite this Biorxiv paper in association with this dataset, the Bibtex for the associated paper is below:

@article {Massenburg2026.06.08.731030,
    author = {Massenburg, Lynnicia N. and Madugula, Sita S. and Brown, Spenser R. and Bible, Amber N. and Harris, Chanda R. and Retterer, Scott T. and Morrell-Falvey, Jennifer L. and Vasudevan, Rama K. and Williams, Alexis N.},
    title = {TileBac: A Benchmark CryoEM Dataset of Bacteria in Ultralow-Dose Montage Tiles},
    elocation-id = {2026.06.08.731030},
    year = {2026},
    doi = {10.64898/2026.06.08.731030},
    publisher = {Cold Spring Harbor Laboratory},
    abstract = {Current segmentation models are capable of routine identification of biological features in noisy cryogenic electron microscopy (cryoEM) images. However, there are still challenges with complete segmentation of high boundary, thin objects such as bacterial cell envelopes and flagella. Moreover, ultralow-dose cryoEM images pose as an additional challenge to boundary distinctions between the object and background. Here, we present TileBac, a benchmark dataset of ultralow-dose montage tiles of Pantoea sp. YR343 to segment bacterial inner and outer membranes for evaluation of model effectiveness. We show that foundation models outperform convolutional neural networks at continuous bacterial cell envelope segmentation despite having lower performance metrics. We release the TileBac benchmark dataset on Hugging Face for further insights into model architecture development.Competing Interest StatementThe authors have declared no competing interest.U.S. Department of Energy, Office of Science, https://ror.org/00mmn6b08, FWP ERKCZ64UT-Battelle, LLC, https://ror.org/04nza6677, DE-AC05- 00OR22725Oak Ridge National Laboratory, https://ror.org/01qz5mb56},
    URL = {https://www.biorxiv.org/content/early/2026/06/09/2026.06.08.731030},
    eprint = {https://www.biorxiv.org/content/early/2026/06/09/2026.06.08.731030.full.pdf},
    journal = {bioRxiv}
}

This readme file was generated on 2025-05-28 by Lynnicia Massenburg

GENERAL INFORMATION

  1. Title of Dataset: tilebac-flag-dataset

  2. Author Information A. Principal Investigator Contact Information Name: Alexis Williams ORCID: 0000-0002-5283-5822 Institution: Oak Ridge National Laboratory Email: williamsan@ornl.gov B. Alternate Contact Information Name: Lynnicia Massenburg ORCID: 0000-0002-6590-273X Institution: Oak Ridge National Laboratory Email: massenb2@hotmail.com

  3. Date of data collection: 2025-06-26

  4. Geographic location of data collection: Oak Ridge, TN

  5. Information about funding sources that supported the collection of the data:

This work is supported by the U.S. Department of Energy, Office of Science FWP ERKCZ64, Structure Guided Design of Materials to Optimize the Abiotic-Biotic Material Interface, as part of the Biopreparedness Research Virtual Environment (BRaVE) initiative. Sample preparation, imaging and image analysis were conducted as part of a user project at the Center for Nanophase Materials Sciences (CNMS), which is a US Department of Energy, Office of Science User Facility at Oak Ridge National Laboratory. Electron microscopy data was collected using instrumentation within ORNL's Materials Characterization Core provided by UT-Battelle, LLC, under Contract No. DE-AC05- 00OR22725 with the DOE and sponsored by the Laboratory Directed Research and Development Program of Oak Ridge National Laboratory, managed by UT-Battelle, LLC, for the U.S. Department of Energy.

SHARING/ACCESS INFORMATION

  1. Reuse restrictions placed on the data: MIT license

  2. Links to publications that cite or use the data: (paper publication in progress)

  3. Links to other publicly accessible locations of the data: N/A

  4. Links/relationships to ancillary data sets:

All datasets are openly available on Constellation (DOI: 10.13139/ORNLNCCS/3025229).

Datasets and raw cryoEM images can be found on https://github.com/Lynnicia/TileBac-ultralow-dose-montage-tiles as well as on Hugging Face at: https://huggingface.co/datasets/LynnMass/tilebac-ULDM-benchmark-dataset, https://huggingface.co/datasets/LynnMass/tilebac-ULDM-tiles, https://huggingface.co/datasets/LynnMass/tilebac-stitched-montage and https://huggingface.co/datasets/LynnMass/tilebac-flag-dataset.

Models can be found on Hugging Face at: https://huggingface.co/LynnMass/640-SAM3-ULDM, https://huggingface.co/LynnMass/1024-SAM3-ULDM, https://huggingface.co/LynnMass/1024-YOLO26-ULDM, https://huggingface.co/LynnMass/640-YOLO26-ULDM, https://huggingface.co/LynnMass/1024-YOLOv11-ULDM, https://huggingface.co/LynnMass/640-YOLOv11-ULDM, https://huggingface.co/LynnMass/1024-U-Net-ULDM, https://huggingface.co/LynnMass/640-U-Net-ULDM, https://huggingface.co/LynnMass/1024-Detectron2-ULDM and https://huggingface.co/LynnMass/640-Detectron2-ULDM.

  1. Was data derived from another source? If yes, list source(s): No

  2. Recommended citation for this dataset:

Massenburg, L. N., Madugula, S. S., Brown, S. R., Bible, A. N., Retterer, S.T., Morrell-Falvey, J.L., Vasudevan, R. K. and Williams, A. (2026). TileBac: A Benchmark CryoEM Dataset of Bacteria in Ultralow-Dose Montage Tiles. (Submitted to NeurIPS).

  • = joint first author

DATA & FILE OVERVIEW

  1. File List:
~DATASET: tilebac-flag-dataset~ 

FOLDER: Annotated
    SUBFOLDER: COCO
        SUBFOLDER: 1024-2x-train
            SUBFOLDER: test
                FILES: [images].jpg (31)
                FILE: _annotations.coco.json
            SUBFOLDER: train
                FILES: [images].jpg (438)
                FILE: _annotations.coco.json
            SUBFOLDER: valid
                FILES: [images].jpg (28)
                FILE: _annotations.coco.json
            FILE: README.dataset.txt
            FILE: README.roboflow.txt
        SUBFOLDER: 1024-3x-train
            SUBFOLDER: test
                FILES: [images].jpg (31)
                FILE: _annotations.coco.json
            SUBFOLDER: train
                FILES: [images].jpg (657)
                FILE: _annotations.coco.json
            SUBFOLDER: valid
                FILES: [images].jpg (28)
                FILE: _annotations.coco.json
            FILE: README.dataset.txt
            FILE: README.roboflow.txt
        SUBFOLDER: 4096
            SUBFOLDER: test
                FILES: [images].jpg (31)
                FILE: _annotations.coco.json
            SUBFOLDER: train
                FILES: [images].jpg (219)
                FILE: _annotations.coco.json
            SUBFOLDER: valid
                FILES: [images].jpg (28)
                FILE: _annotations.coco.json
            FILE: README.dataset.txt
            FILE: README.roboflow.txt
    SUBFOLDER: YOLO
        SUBFOLDER: 1024-2x-train
            SUBFOLDER: test
                SUBFOLDER: images
                    FILES: [images].jpg (31)
                SUBFOLDER: labels
                    FILES: [images].txt (31)
            SUBFOLDER: train
                SUBFOLDER: images
                    FILES: [images].jpg (438)
                SUBFOLDER: labels
                    FILES: [images].txt (438)
            SUBFOLDER: valid
                SUBFOLDER: images
                    FILES: [images].jpg (28)
                SUBFOLDER: labels
                    FILES: [images].txt (28)
            FILE: README.dataset.txt
            FILE: README.roboflow.txt
            FILE: data.yaml
        SUBFOLDER: 1024-3x-train
            SUBFOLDER: test
                SUBFOLDER: images
                    FILES: [images].jpg (31)
                SUBFOLDER: labels
                    FILES: [images].txt (31)
            SUBFOLDER: train
                SUBFOLDER: images
                    FILES: [images].jpg (657)
                SUBFOLDER: labels
                    FILES: [images].txt (657)
            SUBFOLDER: valid
                SUBFOLDER: images
                    FILES: [images].jpg (28)
                SUBFOLDER: labels
                    FILES: [images].txt (28)
            FILE: README.dataset.txt
            FILE: README.roboflow.txt
            FILE: data.yaml
        SUBFOLDER: 4096
            SUBFOLDER: test
                SUBFOLDER: images
                    FILES: [images].jpg (31)
                SUBFOLDER: labels
                    FILES: [images].txt (31)
            SUBFOLDER: train
                SUBFOLDER: images
                    FILES: [images].jpg (219)
                SUBFOLDER: labels
                    FILES: [images].txt (219)
            SUBFOLDER: valid
                SUBFOLDER: images
                    FILES: [images].jpg (28)
                SUBFOLDER: labels
                    FILES: [images].txt (28)
            FILE: README.dataset.txt
            FILE: README.roboflow.txt
            FILE: data.yaml
FOLDER: Raw
    SUBFOLDER: 
        SUBFOLDER: raw_mrc_avg
            SUBFOLDER: MOPSglu
                FILES: [images].mrc (44)
            SUBFOLDER: MOPSsuc
                FILES: [images].mrc (226)
            SUBFOLDER: R2A
                FILES: [images].mrc (35)
        SUBFOLDER: raw_mrc_fractions
            SUBFOLDER: MOPSglu
                FILES: [images].mrc (44)
            SUBFOLDER: MOPSsuc
                FILES: [images].mrc (226)
            SUBFOLDER: R2A
                FILES: [images].mrc (35)
        SUBFOLDER: raw_mrc_sum
            SUBFOLDER: MOPSglu
                FILES: [images].mrc (44)
            SUBFOLDER: MOPSsuc
                FILES: [images].mrc (226)
            SUBFOLDER: R2A
                FILES: [images].mrc (35)
                
FILE: .gitattributes

FILE: README.md

  1. Relationship between files: Cryogenic electron miroscopy images of Pantoea sp. YR343 bacteria in [images]. Raw images (.jpeg files) and annotated (YOLO text and COCO json files) images. Instances listed in this file are model segmentation predictions per class.

  2. Additional related data collected that was not included in the current data package: No

  3. Are there multiple versions of this dataset? If yes, what files were updated and why? Only one version of the dataset is available.

METHODOLOGICAL INFORMATION

  1. Description of methods used for collection/generation of data: Madugula, S. S., Massenburg, L. N., Brown, S. R., Bible, A. N., Harris, C. R., Zhang, L. X., Parker, K., Retterer, S. T., Morrell-Falvey, J. L., Vasudevan, R. K., & Williams, A. N. (2026). Automated Bacterial Identification and Morphological Feature Analysis in Low-Dose Cryo-EM Using YOLOv11. Advanced Intelligent Discovery, n/a(n/a), e202500241. https://doi.org/https://doi.org/10.1002/aidi.202500241

  2. Methods for processing the data: Annotated data show flagella and bacteria annotations with 2x or 3x image expansion in the train subfolder. Raw original images are included in this dataset.

  3. Instrument- or software-specific information needed to interpret the data: YOLOv11, YOLO26, U-Net, Detectron2, SAM3

  4. Standards and calibration information, if appropriate: N/A

  5. Environmental/experimental conditions: Manual cryoEM image collection was described in Madugula et al. for low-dose images collected at 40 e⁻/Å2 at –5 μm defocus using the Falcon 3EC direct electron detector (Thermo Fisher Scientific, counting mode) on the Thermo Fisher Scientific Krios G4 operated in nanoprobe TEM mode (Madugula et al., 2026).

  6. Describe any quality-assurance procedures performed on the data: visual check

  7. People involved with sample collection, processing, analysis and/or submission: Massenburg, L. N., Madugula, S. S., Brown, S. R., Bible, A. N., Zhang, L., Parker, K., Retterer, S.T., Morrell-Falvey, J.L., Vasudevan, R. K. and Williams, A.

DATA-SPECIFIC INFORMATION FOR: All [images].jpg

  1. Number of variables: 2 classes (class 0: flagella), class 1: bacteria)

  2. Number of cases/rows: 31 valid images

  3. Variable List: (test images) class 0: 67 instances class 1: 20 instances

  4. Codes used for missing data: N/A

  5. Specialized formats or other abbreviations used: N/A
Downloads last month
579
Number of rows:
5,408
Total file size:
454 GB