(P06-019-25) Integrating Machine Learning Into Nutritional Assessment and Food Waste Monitoring in Food Services: A Step Toward Sustainability

Objectives: Food waste and accurate dietary assessment are significant challenges in food services, with substantial implications for sustainability and public health nutrition. The project INTENTION aims to develop and validate a machine learning-based system to recognize food items and estimate consumption and waste from tray images taken before and after meals.

Methods: A machine learning model was identified to recognize food items and estimate portion sizes using images of trays captured before and after meals. An experimental study was conducted involving five participants under controlled conditions. Participants selected and consumed meals under supervision, with tray images and corresponding weights recorded to validate the system's performance in food recognition and consumption estimation.

Results: Preliminary results from the controlled experiments showed that the model achieved over 90% accuracy in food recognition and demonstrated a mean absolute error below 10% for portion size estimation. These findings suggest the model’s effectiveness in identifying food items and quantifying consumption, even in its initial stages of development.

Conclusions: Preliminary results support the feasibility of automated systems for food recognition and consumption estimation in food service settings. Further refinements and validations are underway to improve the model's accuracy and test its application in real-world environments. This approach can potentially enhance food waste monitoring, dietary assessment, and sustainability practices in institutional food services.

Funding Sources: The 'INTENTION' project was funded under the second edition of the Public Notice for the selection of project proposals within Spoke 7 'Policy, Behaviour and Education' of the Extended Partnership ONFOODS, Research and Innovation Program PE_00000003, funded by the Italian National Recovery and Resilience Plan (PNRR)