Dr. Santos Gracia Villar, president of the Ibero-American University Foundation (FUNIBER), highlights an international scientific study that analyzes the use of artificial intelligence to detect diseases in tomato leaves with high precision, an advance that could significantly improve crop health management and contribute to more efficient and sustainable agriculture. The study, published in the scientific journal Sensors, demonstrates how deep learning models can identify plant pathologies at an early stage through automated image analysis.
Tomatoes are one of the most important crops worldwide, both for their economic value and their presence in the diets of millions of people. However, their production is often affected by numerous diseases that damage the leaves of the plant and reduce both productivity and fruit quality. Detecting these diseases at an early stage is essential to avoid significant economic losses and improve food security.
Traditionally, plant diseases have been identified through direct observation by farmers or plant pathologists. Although this method is still widely used, it has significant limitations. These include the possibility of human error, the difficulty of detecting symptoms in the early stages, and the lack of access to experts in many agricultural regions of the world.
Given this situation, in recent years the scientific community has begun to explore the potential of digital technologies to support agricultural diagnosis. The development of tools based on machine learning and computer vision has made it possible to create systems capable of analyzing crop images and recognizing patterns associated with plant diseases.
In this context, the study proposes a model based on convolutional neural networks (CNN), a type of artificial intelligence algorithm specially designed to process images. These networks are capable of identifying complex visual characteristics, such as color changes, spots, irregular edges, or texture patterns that often indicate the presence of pathogens on leaves.
To develop and evaluate the system, the researchers used a database of approximately 3,000 images of tomato leaves, including both healthy leaves and samples affected by different diseases. Before training the model, the images underwent a preprocessing process that improved their quality and isolated the relevant areas of the leaves, thus facilitating the algorithm’s learning.
The results obtained were particularly noteworthy. The developed system achieved an accuracy of close to 98.5% in disease classification, demonstrating the enormous potential of artificial intelligence as a support tool for agricultural diagnosis. This level of accuracy indicates that deep learning models can become strategic allies for the early detection of pathologies affecting crops.
Such tools could facilitate faster and more accurate decision-making, allowing appropriate treatments to be applied at early stages and reducing both production losses and the excessive use of plant protection products.
The study is also part of the growing trend toward smart agriculture, a model that integrates sensors, artificial intelligence, data analysis, and digital technologies to optimize agricultural production and improve environmental sustainability.
The researchers conclude that deep neural networks represent a key tool for the future of precision agriculture. Their ability to analyze large volumes of visual data and detect complex patterns opens up new opportunities for automating crop monitoring, disease prevention, and improving agricultural productivity.
If you would like to learn more about this study, click here.
To read more research, consult the UNEATLANTICO repository.