A novel system for assessing kiwifruit water status by AI and 3D image analysis

G. Tacconi, M. Tonon, P. Marcuzzo, N. Belfiore, M. Minervini, L. Rigato, D. Vicino, N. Vicino, F. Gaiotti
This study aims to develop a novel system for precision irrigation in kiwifruit orchards to face the challenge of water conservation, improve fruit quality and avoid soil-borne diseases. This technology is based on a stereo vision system that collects 3D pictures of the canopy with an integrated infrared thermographic camera. Through advanced image analysis the system allows to monitor the volume/leaf area of the canopy, changes in leaf inclination and in canopy temperature. These measures are dependent on the plant water status and are used in combination with climate and soil moisture data collected from other sensors to assess the water status and requirements of the plants. The field trials were set up in 2020 in four commercial orchards in northern Italy, with two treatments: reduced irrigation (RI) managed to impose a progressive stress; normal irrigation (NI) managed following the conventional irrigation schedules of the area. In-field measurements, including midday stem water potential (SWP) and leaf inclination were performed during the season. The canopy images collected by the camera were used to train artificial intelligence (AI) to detect the canopy and automatically estimate the volume/leaf area, the leaf angle distribution and the average leaf temperature. Preliminary results demonstrate that the system is able to correctly detect the canopy and to record variations in temperature and in leaf inclination. Correlation of these data with the in-field measurements of water status is currently under study: the leaf inclination seems to be a good indicator of water stress in kiwifruit plants.
Tacconi, G., Tonon, M., Marcuzzo, P., Belfiore, N., Minervini, M., Rigato, L., Vicino, D., Vicino, N. and Gaiotti, F. (2022). A novel system for assessing kiwifruit water status by AI and 3D image analysis. Acta Hortic. 1332, 231-238
DOI: 10.17660/ActaHortic.2022.1332.31
https://doi.org/10.17660/ActaHortic.2022.1332.31
optimization, irrigation, 3D image analysis, artificial intelligence AI, fruit quality
English

Acta Horticulturae