Leonardo de Souza

Universidade Estadual de Campinas (UNICAMP)

Papers

Total Citations

H-Index

About

Leonardo de Souza is a leading researcher in agricultural computer vision, with a focused expertise in deep learning for precision viticulture. His most significant contribution is the creation of the **Embrapa Wine Grape Instance Segmentation Dataset (Embrapa WGISD)**, a landmark resource that has become a cornerstone for the field. This meticulously curated dataset, cited over 30 times collectively, provides high-resolution, annotated images of wine grapes, enabling the development and benchmarking of state-of-the-art instance segmentation models. By following the rigorous "datasheet for datasets" framework, de Souza ensured his work is not only technically robust but also ethically and methodologically transparent, setting a new standard for dataset creation in agricultural AI. His work directly addresses the critical challenge of automating yield estimation and disease detection in vineyards, bridging the gap between computer vision research and practical agricultural needs. Through this foundational dataset, Leonardo de Souza has empowered a global community of researchers to advance smart farming, making him a pivotal figure in the application of AI to sustainable agriculture.