Pollen-ITA: Polygon-annotated light-microscopy images of airborne pollen and fungal spores from Italian aerobiological monitoring

Pollen-ITA is a collection of light-microscopy images of airborne pollen grains and fungal spores, with polygon (instance) annotations in COCO format. The images derive from the aerobiological monitoring stations of ARPA Marche (Italian POLLnet network) and from a reference pollen collection held by ARPA Marche. The dataset is intended for training and evaluating segmentation, object detection and classification models.

CONTENTS
- 1,178 RGB images in PNG format, 40x objective.
- 8,867 annotated particles in 25 classes (20 pollen types, 5 fungal spore types), drawn as polygons.
- A single annotation file in COCO format (_annotations.coco.json).

ARCHIVE STRUCTURE
Pollen-ITA_rel201026_1178img/
  train/
    <image files>.png
    _annotations.coco.json

All images are in a single folder named "train", a name inherited from the annotation platform's export format. The dataset is NOT divided into training, validation and test sets. Several images may come from the same slide: when defining your own partitions, keep images of the same slide in the same subset to avoid information leakage.

HOW TO CITE
Please cite the dataset and the associated article:

Dataset: L. Nepi, N. Trobiani, F. Olivieri, E.M. Lodolini, A. Mancini, Pollen-ITA: Polygon-annotated light-microscopy images of airborne pollen and fungal spores from Italian aerobiological monitoring, Zenodo, v2.0, 2026.  https://doi.org/10.5281/zenodo.23511322