REDUCE
AI-driven Solution to Reduce Canteen Leftovers Through User Engagement
Overview
The challenge
In canteens across Europe, nearly one in two trays is returned with substantial amounts of uneaten food — often discarded without any chance of recovery. This results in over 300,000 tonnes of food waste annually, enough to feed a city the size of Oslo three times a day for more than a year. Beyond the ethical and social implications, this waste places a heavy burden on the environment and public budgets.
Tackling this issue requires more than awareness — it demands actionable, data-driven solutions that can reshape food consumption habits and operational efficiency in collective dining spaces.
Users/Testers
30
The solution
Our solution — REDUCE (Reduction through Engagement and Data Utilization for Controlling Excess) — combines advanced artificial intelligence with active user participation to effectively address plate waste in canteens.
REDUCE operates by collecting data across three key domains:
- Food waste quantities and types
- Environmental conditions (temperature, humidity, noise)
- User behaviour and feedback before and after meals
This multidimensional dataset feeds an AI model trained to identify patterns and predict waste under specific conditions. Based on these insights, the system generates data-driven recommendations — such as adjusting portion sizes, menu composition, or serving strategies — enabling catering providers to make smarter, more sustainable choices.
The project is structured around three integrated pillars, each developed by a specialised partner:
- User engagement: Led by Behavix srl SB, this component gathers valuable input directly from canteen users before and after their meals. The goal is to understand satisfaction levels and the behavioural causes of plate waste while promoting a cultural shift toward mindful consumption.
- Food waste recognition: Developed by the TESAF Department at the University of Padova, this custom-built Computer Vision system automatically scans trays to identify and quantify leftovers. This enables consistent, real-time monitoring of waste patterns and serves as a critical input for evaluating intervention outcomes.
- Data analytics and forecasting: Managed by Spritzmatter srl, a university spinoff specialising in AI, this layer applies advanced algorithms to analyse the collected data, predict future waste levels, and prioritise tailored interventions to minimise food waste over time.
REDUCE engages three main stakeholder groups within the canteen ecosystem:
- Canteen end users (employees or diners) provide feedback on their meals and help monitor their own waste, playing an active role in fostering more sustainable habits.
- Canteen service providers use REDUCE to collect user feedback, monitor waste levels, and gain actionable insights to improve service and menu planning.
- Canteen hosts (typically the companies housing the canteens) monitor the overall impact and effectiveness of the service through high-level statistics and dashboards.
A real-world demonstration of REDUCE will take place in a corporate canteen, where meals are served daily to employees. This pilot will validate the effectiveness of AI-enhanced monitoring and behaviour-driven change in reducing plate waste.
By uniting AI technology with stakeholder collaboration, REDUCE offers a scalable, evidence-based strategy to tackle food waste and promote a more sustainable food service model.
Users/Testers
30
Members
Italy
Italy
Italy
Citizen Engagement
Citizens play a central role in the REDUCE project, actively contributing to both the design and testing phases of the solution. To ensure the system was aligned with user needs, we organised focus groups before the pilot phase. These sessions allowed us to present the REDUCE concept, gather early feedback, and co-design specific features related to user engagement and meal evaluation. Insights from these discussions directly influenced the structure of pre- and post-meal feedback interactions and the design of awareness campaigns.
During the pilot, held in a corporate canteen, users are involved daily by sharing feedback before and after meals — commenting on satisfaction, portion sizes, and reasons for food waste. These subjective inputs are combined with data collected by a computer vision system that scans trays to measure leftovers objectively. This integration enables REDUCE to train AI models that deliver targeted, actionable recommendations to reduce plate waste.
In addition to meal-based interactions, citizens are engaged through behavioural nudges and awareness initiatives developed by Behavix srl SB to encourage sustainable consumption habits
DataU and FOODITY components
The FOODITY DataU is integrated into the REDUCE solution to ensure GDPR compliance and promote true data sovereignty. While personal data remains stored securely within REDUCE, DataU provides users with a decentralised mechanism to manage their data rights. This means users have full visibility and control over the data collected about them, including the ability to decide if, when, and with whom it can be shared. If users choose, they can authorise the sharing of their data with third parties — such as other companies or services — through DataU’s secure and consent-based infrastructure. This approach not only builds trust and transparency but also ensures compatibility with broader FOODITY components and future data ecosystems.
Materials and links
Alignment with FOODITY
REDUCE is strongly aligned with FOODITY’s key objectives, particularly in promoting data sovereignty, citizen engagement, food ethics, and transparency.
By integrating DataU, REDUCE ensures that all personal data remains under the control of the users, who can manage access and sharing through a secure, decentralised infrastructure. This guarantees GDPR compliance while empowering citizens to retain ownership of their data.
The solution also places citizens at the centre of the process through continuous engagement activities, such as feedback collection, awareness campaigns, and focus groups, encouraging behavioural change and co-creation.
Ethically, REDUCE addresses the pressing issue of food waste in a socially responsible manner, helping canteens reduce environmental impact while respecting user preferences and dietary habits. Throughout the process, transparency is maintained in both data usage and intervention strategies, ensuring trust and accountability among all stakeholders involved.
Lessons and recommendations
Lessons learned
A key lesson learned from the development and deployment of REDUCE is the complexity of communicating and coordinating with companies during the pilot phase. Bureaucratic processes and detailed requirements often slow progress, making timely execution more challenging. Additionally, installing and integrating systems directly within the canteen environment adds an extra layer of overhead, with many unforeseen edge cases that can only be addressed through hands-on experience. These challenges have highlighted the importance of flexibility and adaptability in managing real-world deployments, where unexpected situations often arise and require quick problem-solving.
Recommendations
A key recommendation for replication or scalability is to prioritise security-by-design when developing solutions that handle GDPR-protected data. This approach ensures that data privacy and security are integrated into the system from the very beginning, rather than being added as an afterthought. While this may incur higher initial costs, it ultimately simplifies the process and reduces long-term challenges related to compliance, data breaches, and privacy concerns. By establishing strong security foundations early on, developers can ensure a scalable and sustainable solution that respects data rights and fosters trust with users, making it easier to expand and adapt to new contexts in the future.
Main benefit of participating in the FOODITY Programme
The main benefit of participating in the FOODITY Programme is access to mentoring sessions that provide invaluable support across key areas of the project. From development and user engagement to market strategy, these sessions offer tailored guidance to refine and optimise every aspect of REDUCE.
FOODITY’s expertise not only helps enhance the technical and operational elements but also ensures the solution aligns with data sovereignty and food ethics principles. Additionally, being part of a community of like-minded projects that share values and goals fosters a collaborative environment that encourages knowledge sharing and mutual growth. This collective synergy accelerates our progress, enhances our solution’s impact, and provides opportunities for scaling within the broader food sector.