Abstract

Food waste (FW) and unsustainable food consumption are linked to several critical issues, including food poverty, food insecurity, environmental crises, and economic losses. Despite decades of research and the implementation of mitigation strategies by policymakers, industry leaders, and academic experts, consumer-centric FW remains under-studied due to challenges such as the lack of empirical data. The present study, part of the H2020 LOWINFOOD project, adopts a consumer-centric approach by utilizing primary data collected through the Regusto Innovation and Regusto bags to quantify FW generated both out-of-home and brought home. The study aims to measure consumer FW and evaluate the effectiveness of food-sharing platforms for waste mitigation, with a particular emphasis on the post-consumption stage. Additionally, innovative AI-computer vision technologies were employed to verify the accuracy of FW measurements (in weight) collected via consumer surveys, thus addressing the “mystery” of consumer post-consumption FW and enhancing the evaluation of the waste-mitigating digital tool.
The results indicate a significant reduction in FW by 88.16% and substantial consumer engagement, underscoring the effectiveness of the innovation. The AI-computer vision solution achieved 98% accuracy in food recognition. However, a notable discrepancy of approximately 50% was observed between self-reported and AI-estimated FW weights, suggesting that FW quantification based on consumer surveys is likely overestimated. Furthermore, the findings reveal variations in dining preferences and socio-demographic factors between consumers who take away leftovers and those who order meals from food surplus. These insights are valuable for policymakers and business innovators seeking to mitigate FW by tailoring strategies to specific restaurant types and household dynamics.