The information website by, for and about
the tomato processing industry globally

Israel: machine learning meets tomato farming

10/02/2025

François-Xavier Branthôme
Israel,
Middle East
${printContents} `); printWindow.document.close(); printWindow.focus(); printWindow.print(); printWindow.close(); }); });
Israeli researchers have developed a machine learning model that uses hyperspectral imaging to assess the quality of tomatoes before harvest. It is a cost-effective and non-destructive method for predicting critical quality parameters, including weight, firmness, and lycopene content. This innovative approach allows farmers to monitor fruit development in real time, optimize harvest timing, and improve the overall quality of harvested fruit. The results of this research represent a significant leap forward in precision agriculture and sustainable food production.

 Dr. David Helman

A research team led by Dr. David Helman from the Faculty of Agriculture, Food, and Environment at the Hebrew University of Jerusalem has developed a new machine learning model that uses hyperspectral imaging to assess the quality of tomatoes before harvest. Hyperspectral images of specific wavelengths of light, called spectral bands, are used to study the properties of fruits based on how they reflect light. This innovative new approach addresses challenges posed by traditional methods and is faster, non-destructive, and more cost-effective.

“Traditional methods of assessing the quality of tomatoes are time-consuming, destructive and costly, limiting their use to small samples after harvest. Farmers lack practical tools to monitor quality before harvest, which is essential for optimizing growing conditions and improving crop quality,” the lead author of the study and director of the Hebrew University of Jerusalem’s Vegetation Systems Modelling and Monitoring Laboratory, Dr. David Helman, who conducted the study, told the news magazine Food Ingredients First.

The study, conducted in collaboration with researchers from Bar-Ilan University and the Volcani Center, used a handheld hyperspectral camera to collect data on 567 tomatoes from five cultivars. The researchers used machine learning algorithms, including Random Forest and Artificial Neural Networks, to predict seven critical quality parameters: weight, firmness, total soluble solids (TSS), citric acid, ascorbic acid, lycopene, and pH. The models demonstrated high accuracy, with the Random Forest algorithm achieving an R² of 0.94 for weight and 0.89 for firmness, among others.

The study mainly led to the following conclusions:
• Effectiveness in the choice of spectral bands: a model with five spectral bands is sufficient to effectively predict the determined quality parameters, which makes it possible to develop less expensive portable devices;
• Flexibility of implementation: the model proves to be both robust and scalable regardless of varieties and growing conditions;
• Gain in terms of harvest: precise monitoring of the state of ripening and quality of fruits allows producers to optimize the choice of harvest date and thus improve the quality of harvested products.
 

 ToMAI-SENS imaging of fruits at different bands, identifying the fruit and estimating its quality parameters.
"In addition to improving nutritional quality, the team’s AI-based technology could also enable “better adaptation to environmental changes, strengthen the resilience of agricultural systems and contribute to global food security,” explained Dr. Helman.
“Our research aims to bridge the gap between advanced imaging technology, AI, and practical agricultural applications,” he adds. “This work has the potential to revolutionize quality monitoring not only in tomatoes but also in other crops. Our next step is to build a low-cost device (ToMAI-SENS) based on our model that will be used across the fruit value chain, from farms to consumers.”

The study highlights the opportunities created by integrating this technology into agricultural practices, from intelligent harvest optimization systems for professionals to tools that can be used by the general public to assess the quality of produce in supermarkets.

The research paper, “Hyperspectral imaging-based machine learning models for monitoring tomato fruit quality before harvest,” is available in Computers and Electronics in Agriculture and can be found at https://doi.org/10.1016/j.compag.2024.109788. 

Sources: cfhu.org, sciencedirect.com, phys.org, foodingredientsfirst.com