Tuberculosis Bacilli Detection Dataset: Ziehl–Neelsen Stained Sputum Smear Microscopic Images with Expert Annotations

This dataset contains 101 full-field microscopic images of Ziehl–Neelsen stained sputum smear samples and their corresponding expert-annotated ground truth images for tuberculosis bacilli detection.

Each raw image is paired with a corresponding ground truth image of the same filename. Ground truth annotations are provided as visual overlays, where circles indicate isolated bacilli and rectangles/squares indicate bacilli clusters.

The following items must be cited if using this dataset:

1. 

@dataset{greeshma_tb_dataset_2026,
  author       = {Greeshma K and Vishnukumar S},
  title        = {Tuberculosis Bacilli Detection Dataset: Ziehl–Neelsen Stained Sputum Smear Microscopic Images with Expert Annotations},
  year         = {2026},
  publisher    = {Zenodo},
  doi          = {10.5281/zenodo.19280029},
  url          = {https://doi.org/10.5281/zenodo.19280029}
}

2.  The article is under consideration at "The Visual Computer", Springer journal titled as "Tuberculosis Bacilli Detection Enhanced: A Hybrid Attention Residual U-Net and Vision Transformer Approach".