Smart olive growing: the ResOILent project for the valorization of genetic resources

A new frontier of Agriculture 4.0 and Precision farming is reached with the ResOILent project, which is part of the Horizon 2020 Gen4Olive.

The aim of Gen4Olive is to optimize the use of olive genetic resources and enhance pre-breeding. The olive grower or technician will be able to identify easily some crop characteristics including:

• resistance to pests and diseases
• plant productivity
• resistance to climate change.

ResOILent is a two-year project funded by Gen4Olive and coordinated by Agricolus, which also involves the Spanish company BrioAgro Technologies.

Given the importance of the production and consumption of olive oil in the Mediterranean area, the objective of ResOILent is to improve the selection process of olive cultivars, promoting the creation of drought-tolerant varieties that are better adapted to different geographical areas.

The project focuses on data acquisition and the identification of certain typical olive characteristics, in order to reduce time and cost of the breeding process. This is possible through the development of artificial intelligence algorithms capable of identifying differences in the phenotype of olive varieties via smartphones.

In the first six months, the aim was to “teach” the model to recognize the chosen characteristic, in our case the leaves, within a photograph. This is the most complex step in the process, also in terms of timing, as a lot of initial data is required to obtain good results.

To understand the complexity of this step, you have to think that within the project, input datasets containing around 5000 leaves were used and each of these was labelled allowing the artificial intelligence algorithm to recognize and discriminate it from its surroundings.

In the following image (figure 1), the first results of the algorithm can be seen: on the left is the input photograph, and on the right the output of the model. The results of the model are very promising as the algorithm was able to identify most of the leaves in the image.

Figure 1 represents only the first step towards the development of the algorithm, which will be improved and refined during the course of the project.

ResOILent represents an excellent opportunity to put into practice the Agricolus technical team’s knowledge of machine learning and deep learning by taking a further step towards Agriculture 4.0.