Dianyo/fecal-health

PCR-Annotated Poultry Health Fecal Image Dataset (4-Class)

Dataset Summary

This repository provides a Hugging Face-compatible version of the "Machine Learning Dataset for Poultry Diseases Diagnostics - PCR annotated".

The goal of this Hugging Face port is to make this highly rigorous dataset instantly accessible for machine learning pipelines, specifically for training PyTorch models, Vision Transformers (ViTs), and simulating Federated Learning (FL) environments.

What makes this dataset exceptional is its ground-truth accuracy. Unlike many visual datasets, the labels here were confirmed via Polymerase Chain Reaction (PCR) molecular diagnostics at a laboratory level. The dataset contains 1,255 high-quality, PCR-verified fecal images collected from poultry farms in Tanzania.

The images are categorized into four distinct classes:

• Coccidiosis (Cocci): 373 images
• Healthy: 347 images
• Salmonella: 349 images
• Newcastle Disease (NCD): 186 images

Because of the strict PCR verification and the real-world, noisy backgrounds of the images, this dataset is an ideal benchmark for multi-class avian disease detection and testing decentralized AI architectures.

How to Use

You can easily load this dataset using the Hugging Face datasets library:

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("Dianyo/fecal-health")

Original Source & Attribution

I did not collect or sequence these images; I have organized and formatted them for the Hugging Face ecosystem to enable easier ML pipeline integration. All credit for the rigorous data collection and PCR laboratory work goes to the original authors.

• Original Dataset: Machine Learning Dataset for Poultry Diseases Diagnostics - PCR annotated
• Authors: Machuve, Dina; Nwankwo, Ezinne; Lyimo, Emmanuel; Maguo, Evarist; Munisi, Charles (2021)
• Original Host: Zenodo

Citation

If you use this Hugging Face dataset in your research or machine learning pipelines, please cite both the original authors who conducted the PCR diagnostic collection and this Hugging Face repository.

1. Cite the original data collection:

@dataset{machuve_dina_2021_5801834,
  author       = {Machuve, Dina and Nwankwo, Ezinne and Lyimo, Emmanuel and Maguo, Evarist and Munisi, Charles},
  title        = {Machine Learning Dataset for Poultry Diseases Diagnostics - PCR annotated},
  month        = dec,
  year         = 2021,
  publisher    = {Zenodo},
  version      = {3},
  doi          = {10.5281/zenodo.5801834},
  url          = {https://doi.org/10.5281/zenodo.5801834}
}

2. Cite this Hugging Face dataset port:

@misc{dianyo2026fecalhealth,
  author = {Tien-Yu Chi},
  title = {Hugging Face Dataset Port: PCR-Annotated Poultry Fecal Images for 4-Class Disease Detection},
  year = {2026},
  url = {https://huggingface.co/datasets/Dianyo/fecal-health}}
}