This site exploits session and third parties cookies. Pressing on 'Accept' button, you accept cookie usage. For further information click Cookie Policy button

  • Banner

Agrosat+ Barilla

Funded under: Barilla S.p.A.

Start date 1 February 2020 End date 31 January 2022

Keywords: Precision Agriculture, Artificial Intelligence, Deep Learning, Durum wheat

In the framework of precision agriculture, the Agrosat+ project aims at developing methods for the classification of images and videos based on cutting-edge machine learning algorithms. The ultimate goal is to develop a real-time software system for the classification of plants, their diseases, weeds and insects based on images shot by mobile devices in uncontrolled scenarios to support farmers and operators during the daily routine. The precise knowledge of diseases and weeds (also obtained thanks to correlation with other data and computational models) will help farmers choose adaptive and optimal treatment to prevent crop losses.

The role of ISTI is focused on the development of classification modules using Artificial Intelligence (AI), specifically Deep Learning models. The datasets ISTI is working on are mixed: some require multiple identifications in addition to a classification. A plethora of models, divided by the task they have to accomplish and possibly to the part of the plant they refer to, have been trained and validated in durum wheat production. In addition, a comparison of the improvements of our developed models is being performed on the state-of-the-art of other benchmark datasets.

Finally, with the development of interfacing solutions between the mobile App and the Artificial Intelligence module, the workflow of the AI module development has started. A set of load-balancing solutions has been implemented and tested.

In addition to aspects relating to cost reduction, this research will also contribute to improving environmental sustainability and green entrepreneurship.