Food loss and waste are often discussed in terms of what happens to food that is no longer consumed. Yet preventing losses also means intervening earlier, at the stages where quality is assessed, products are classified and decisions are made about what can proceed further along the food chain.
The International Day of Awareness of Food Loss and Waste, observed on 29 September, provides an opportunity to look at this less visible dimension of the challenge. Recognised by the United Nations and led by the FAO and UNEP, the Day promotes awareness and action to reduce food loss and waste, with the 2026 observance focusing on the transformative potential of reducing food loss and waste.
In food processing, artificial intelligence and computer vision are increasingly being explored for automated quality assessment. By analysing visual characteristics and identifying deviations from expected quality parameters, such technologies can support more consistent inspection and quality-control processes. Their potential contribution to reducing food loss, however, depends on something more fundamental: whether these systems can perform reliably under conditions representative of real food-processing environments.
This is where testing and validation become particularly relevant. A solution may perform well during development, but its practical value ultimately depends on how reliably it can assess real products against defined quality criteria. Generating appropriate reference data, testing performance under controlled conditions and identifying limitations before deployment are therefore important steps in moving an AI-based solution closer to operational use.
Within agrifoodTEF, this process is supported through dedicated testing and experimentation services. One example is "Computer Vision for Automated Food Quality Assessment", a service provided by the Universitat de Lleida. It enables providers of existing AI-based food quality assessment solutions to evaluate their performance using a controlled testing infrastructure, including a Food Processing Pilot Plant. Where relevant, reference datasets containing food samples and expected quality measures can also be generated for validation and further training. The testing process typically takes up to two weeks, depending on the solution and the number of samples assessed, and results are documented in a report detailing the system's performance metrics.
Within the agrifoodTEF catalogue, a second service, "Applications of artificial intelligence and visual recognition for assessing the quality of agri-food industry products", provided by the Università degli Studi di Napoli Federico II, approaches the same broader challenge from the perspective of AI-based quality assurance and visual recognition. The service is designed to identify deviations from defined quality standards and support proactive quality control, with detection capabilities that can be tailored to characteristics including size, shape, colour and packaging. By helping identify quality deviations before they translate into further production or supply-chain consequences, such approaches can contribute to more informed decisions concerning agri-food products.
Neither service is presented as a food-loss reduction solution in itself. Their relevance to the challenge lies elsewhere: more reliable and better-validated quality assessment can help create the conditions in which food-processing decisions are made on the basis of more accurate information. In this sense, AI validation becomes part of a broader effort to reduce unnecessary losses and improve the efficiency with which food resources are used.
At Trust-IT Services, supporting projects such as agrifoodTEF means contributing to the wider ecosystem that allows these technologies to progress from development towards practical application. Through communication, dissemination and community-building, Trust-IT helps connect technology providers, researchers, testing facilities and the wider agrifood community, making available knowledge, services and opportunities more accessible to those working to advance digital innovation in the sector.
This role reflects a broader principle: innovation does not end when a technology is developed. It also requires the infrastructures, expertise and collaborative environments that allow solutions to be tested, understood and progressively improved. For AI in food processing, this means creating credible pathways between technological development and the complex conditions in which food products are actually assessed and handled.
Looking ahead
As the agrifood sector continues to explore AI, computer vision and other data-driven technologies, the question will increasingly move from whether these technologies can be developed to how confidently they can be deployed. Testing and experimentation facilities can play an important role in that transition by providing the environments in which performance can be evaluated and evidence can be generated before wider adoption.
Reducing food loss and waste will ultimately require action across the entire food system. Technologies capable of improving how food is assessed, classified and monitored may form one part of that wider effort, provided that their performance can be demonstrated in the environments where they are intended to operate.
For projects such as agrifoodTEF, this is where innovation and impact begin to meet: not simply in developing new AI capabilities, but in creating the conditions for those capabilities to be tested, trusted and put to use.
