Annotated dataset of microscope images of pollen grains in honey from 17 beekeeping taxa

Annotated dataset of microscope images of pollen grains in honey from 17 beekeeping taxa

Melissopalynology is a method based on the separation of pollen grains present in honey and the identification of the plant species to which they belong. It is used to determine the botanical, but also the geographical origin of the honey, as well as its commercial value. For this reason, a database including microscope images and characteristics of pollen grains of 17 beekeeping taxa, usually present in honey samples, was created. 

For the honey preparations the methodology of Louveaux et al. (1978) and Von Der Ohe et al. (2004) was followed. Specifically, 5.0 g of honey were weighed and dissolved in 10 ml of distilled water. The solution was centrifuged for 10 min at 2300 r/min. The supernatant solution was discarded and the precipitate was transferred with a disposable plastic Pasteur pipette onto a slide, where it was spread with the addition of fuchsin on a 22 x 22 mm surface. Staining with fuchsin helps to see in greater detail the morphological characteristics of the pollen grains. The preparation was dried by gentle heating to 40°C, on a heating plate and covered with a coverslip on which a small amount of Entellan adhesive (Merck) has been placed. The pollen grains were photographed on an optical microscope (Olympus SZX12), with lens 40× (Olympus DF PLAPO 1X DF) and a digital analysis camera (Olympus SC30), while a morphometry software (Image Pro Plus Software, V1.1.19) was used for their determination. For the microscopic identification of the pollen types, the collection of reference slides from the Laboratory of Apiculture of the Aristotle University of Thessaloniki, which is accredited to ISO 17025:2017, was used.  

The dataset contains 1404 training captured microscope images of pollen grains from 17 major beekeeping taxa (class list can be found below) and 85 testing captured images. Polygon annotations were created using LabelMe software and saved in COCO Annotation format (train.json and val.json files). 

Further information about the related project (SmartBeeKeep) can be found in the following article and presentation (please site if you use these data):

• Vasilios Liolios, Dimitrios Kanelis, Maria-Anna Rodopoulou, Chrysoula Tananaki (2023). A Comparative Study of Methods Recording Beekeeping Flora. Forests, 14(8), 1677; https://doi.org/10.3390/f14081677 
• Nikos Grammalidis, Andreas Stergioulas, Aggelos Avramidis, Konstantinos Karystinakis, Athanasios Partozis, Athanasios Topaloudis, Georgia Kalantzi, Chrisoula Tananaki, Dimitrios Kanelis, Vasilis Liolios, and Madesis Panagiotis "A smart beekeeping platform based on remote sensing and artificial intelligence", Proc. SPIE 12786, Ninth International Conference on Remote Sensing and Geoinformation of the Environment (RSCy2023), 127860C (21 September 2023); https://doi.org/10.1117/12.2681866 Event: Ninth International Conference on Remote Sensing and Geoinformation of the Environment (RSCy2023), 2023, Ayia Napa, Cyprus Author preprint available

Annotation - Latin name

Myrtus             -           Myrtus communis

Brassicaceae   -           Brassicaceae

Cercis  -           Cercis siliquastrum

Helianthus annuus      -           Helianthus annuus

Lavandula       -           Lavandula angustifolia

Robinia pseudacacia   -           Robinia pseudoacacia

Olea    -           Olea europaea

Citrus  -           Citrus sp.

Paliurus           -           Paliurus spina-christi

Eucalyptus      -           Eucalyptus sp.

Polygonum     -           Polygonum aviculare

Carduus           -           Silybum marianum

Cistus  -           Cistus sp.

thymus -           Thymus sp.

Castanea         -           Castanea sativa

erica    -           Erica manipuliflora

Gossypium      -           Gossypium hirsutum