CucurPollen

A microscopy image dataset of pollen germination and tube growth in four cucurbit species: cucumber, melon, pumpkin, and watermelon. The dataset comprises 3089 images acquired during two independent assays performed under heat stress conditions. Images are organized by assay and species and are accompanied by experimental metadata. A subset of 122 images includes instance-level annotations in COCO format distinguishing non-germinated grains, germinated grains, and pollen tubes. An additional subset of 639 images was visually evaluated by an expert using ordinal scores (0-3) to assess pollen abundance, germination level, and tube length development. Images were acquired using two microscopy systems without post-processing prior to deposition. The dataset captures variability in imaging conditions and pollen morphology, including a small subset of 129 images flagged for visual noise. These characteristics may support applications in quantitative reproductive biology as well as computer vision tasks such as instance and semantic segmentation. The dataset can be used for developing automated methods for pollen analysis and for studying pollen performance under heat stress conditions.