Machine Learning Dataset for Poultry Diseases Diagnostics

The annotated dataset of poultry disease diagnostics for small to medium-scale poultry farmers consists of poultry fecal images. The poultry fecal images were taken in Arusha and Kilimanjaro regions in Tanzania between September 2020 and February 2021 using Open Data Kit (ODK) app on mobile phones. The typical normal fecal material which is the ‘healthy’ class and Coccidiosis disease, the ‘cocci’ class were taken from poultry farms. The chickens were inoculated for Salmonella disease and fecal images taken from the diseased chickens for the ‘salmo’ class after one week. The chickens were also inoculated for Newcastle disease and fecal images for the 'ncd' class were taken within three days.

All images are in the .zip files; “cocci.zip” has 2103 images, “healthy.zip” has 2057 images, “salmo.zip” has 2276 images, "ncd.zip" has 376 images. A total of 6,812 image files are labeled.

The “imgObjDet_Yolo.zip” and “imgSegmentation.zip” files consist of the corresponding annotations for object detection on YOLO framework and JSON files for semantic segmentation tasks respectively.

The research project is funded by the International Development Research Centre (IDRC) with IDRC Grant Number: 109187-002.