Deep Learning Approach for Terrace Vineyards Detection from Google Earth Satellite Imagery

A deep learning model demonstrates promising accuracy in detecting terrace vineyards in the UNESCO-listed Alto Douro Wine Region, Portugal. Using U-net architecture applied to Google Earth Pro satellite imagery, researchers achieved preliminary success in automating the identification of these unique agricultural features on rugged schist slopes. This approach could facilitate multi-temporal crop evaluation and cultural characterization, addressing challenges posed by difficult terrain and manual labor intensity. The study highlights the potential of artificial intelligence in enhancing remote sensing applications for heritage site analysis.

Figueiredo, N., Neto, A., Cunha, A., Sousa, J. J., & Sousa, A. (2022). Deep Learning Approach for Terrace Vineyards Detection from Google Earth Satellite Imagery. IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium, 5824–5827. https://doi.org/10.1109/igarss46834.2022.9884644