2.5. Artificial intelligence and image analysis for breeding and varietal characterization

Olive oil and table olives are key agricultural commodities, and the ability to accurately distinguish between different olive varieties is crucial for plant nurseries, growers, researchers and inspection authorities. Image analysis offers a non-invasive and efficient way to achieve this goal.

One of the primary applications of image analysis in olive variety identification is the analysis of fruit and endocarp characteristics. In fact, endocarp morphological characteristics are more informative than any other part of the olive tree. Each olive variety exhibits distinct features, such as fruit shape, size and colour, as well as differences in the shape, size and texture of their endocarps. Using high-resolution images, specialised software can be used to extract and analyse these features, allowing for the identification of specific olive varieties.

Machine learning algorithms are commonly employed in this process. These algorithms are trained on datasets of labelled olive images, with each image linked to its the corresponding olive variety. Through this training, the algorithm learns to recognise patterns and features unique to each variety. Once trained, it can accurately classify olives based on new images, even if they come from different orchards or regions. Currently, recognition accuracy for unknown samples reaches up to 90%, and significant improvements are expected in the coming years. Some examples of applications for olive variety identification include “Olivar” and “OliveID”, alongside various research methodologies presented in scientific literature. In simple terms, users can download an application onto their mobile devices, take photographs of olive fruits and endocarps, and receive a high-probability suggestion for of the variety name — offering speed, convenience, and cost-effectiveness.

Although full (100%) identification accuracy has not yet been achieved, continued technological advances are expected to further improve performance. Image analysis is anticipated to play an increasingly important role in precise plant nursery management and the sustainable cultivation of olive groves.
One of the most significant applications of this technology is the verification of varietal identity during the certification process of nursery material. Another important application is the early selection of plants with desirable traits during breeding programmes.