The dataset contains labels about the the phenological stages of cotton at the parcel level, generated by in-situ inspections in the region of Orchomenos, Greece, during the cotton cultivation period of 2021 (May-October). We collected 1,285 crop growth ground observations. We introduce a new collection protocol, assigning up to two phenology labels that represent the primary and secondary growth stage in the field and thus indicate when stages are transitioning. More information about the annotation procedure can be found in the relevant publication "Fuzzy clustering for the within-season estimation of cotton phenology". Along with the annotations we are publishing a unique dataset of 3,142 images. Each image is named in the following format: {date}_{id}-{type}.jpg, where "date" is in the format {year}{month}{day} and refer to the exact date the image was captured, "id" is the unique id of a parcel and "type" is one of "O", "A" or "B". "O" refers to a panoramic photo of the entire field, "A" to a photo that is representative of the majority of the plants in the field and "B" to one representative of a minority of plants in the field. For each date and each unique parcel the first two exists always while the latter close-up photo had be captured only when the percentage of the minority class, in terms of area, was deemed significant. If you use this dataset, please cite our paper: @article{sitokonstantinou2022fuzzy, title={Fuzzy clustering for the within-season estimation of cotton phenology}, author={Sitokonstantinou, Vasileios and Koukos, Alkiviadis and Tsoumas, Ilias and Bartsotas, Nikolaos S and Kontoes, Charalampos and Karathanassi, Vassilia}, journal={PLOS ONE}, year={2023}, doi={10.1371/journal.pone.0282364} }