Case Studies

The Bin Project

AI- generated labels for food waste identifying the food wasteUsing AI technology & behavioural economics to minimise food wastage

— Details at a glance
Project Title The Bin Project
Year 2023
Location Singapore
Country Singapore
Genre Technology, Waste
Lead Name(s) Hanming Ye
Other Organisations UWCSEA East and Dover campus
Value $1000
Project Report Download Report

Background

This project aimed to analyse food wastage using AI technology and feedback to the food preparation management where savings could be made.

With the support of GoMADʼs funding, we designed and set up smart cameras at UWCSEA East and Dover campus. Each lunchtime, the smart cameras automatically take pictures of studentsʼ food waste. Then, we were able to create AI models that classify the food waste pictures taken. The food waste was captured by the smart camera at UWCSEA East, with computer-generated labels identifying the dishes as pad thai noodles.

Then, by analyzing the resulting data, we have been able to isolate the problematic dishes and take action accordingly. For instance, by analyzing 3161 plates of food waste over two weeks in October 2023, we realized that the pad thai dish served every Monday accounts for more than 60% of all waste generated by the East campus canteen. In other words, one dish out of a dozen served was behind more than two-thirds of the waste. Based on this finding, we recommended that the canteen reduce portion sizes for pad thai by 15%. The canteen opted to remove pad thai from its menu for the rest of the year instead.

A website was developed to spread awareness to students and other schools about our work (work in progress): thebinproject.org

 

How the Funding Was Used

  • Acquire electronic components
  • Create an eco-friendly camera housing
  • Purchase local and cloud storage
  • Create a website
  • Adapt and purchase technologies

How They Made A Difference

As a result, we observed that food waste decreased by 20-30% on Mondays after this change. You can find more detailed data analyses over this period and our recommendations to the canteen in this document.

A total of 3161 plates were captured across the two weeks from Oct 2nd to 13th. The type and proportion wasted of each plate discarded was identified by a machine learning model, with manual verification. We observed that food waste decreased by 20-30% after this information was used to bring about a change in portion sizes.

 

Our Future Plans

Photos capturing the food waste prior to automatic anaylsing

After testing our prototype over the last year, weʼre looking to create a newer version of the smart camera that can run 24/7 without maintenance to make sure that no plate of food waste goes unaccounted for. Weʼre also learning from our past analyses to improve the accuracy of our AI models.

We believe in the long-term potential of The Bin Project in making food consumption sustainable. As the original four team members have graduated in 2024, we have been working to ensure the continuity of the project. We have reached out to and onboarded six new team members – now our team totals ten members! Ranging from middle school to IB, the expanded team will continue operating the project over the coming years, further its technological development, and spread awareness about our cause to the local community.

How The Project Made A Difference for the Volunteers

This project enabled the students to work with emerging technologies to provide meaningful data to reduce food waste.

Quote

Through the project, we learnt that shaping sustainable behavior is hard, but possible if all stakeholders come together to reflect on our environmental impact.