My thesis project aims at developing an automated system for counting and identifying pollen grains and to apply it to contemporary questions specific to the Mediterranean region.
Image recognition techniques based on Deep Learning approaches will be used.
The developed tool will be applied to modern pollen grains to study the mechanisms structuring pollen-pollinator-plant networks in a polluted environment and the interannual variability of pollination of wild and cultivated grapevines in a context of increased drought. It will also be applied to fossil pollen grains (from sedimentary cores) to study vegetation changes and vegetation/climate/anthropic impact interactions during the Holocene (last 11700 years).