Our first talk of the AIGSA Graduate Student Tutorials will be given by Aidan Beery! More information can be found below.
Title: Garbage In, Knowledge Out: A Smart Compost Bin for Automatic Food Waste Management
Summary: Food waste is a growing crisis, with over 1.3 billion tons of food being wasted each year, while over 783 million people go hungry. At least 60% of this waste occurs at the consumer level, and yet there exists no scalable, repeatable method for measuring household food waste. We present the Smart Compost Bin, an AI-assisted kitchen compost bin built to automatically measure food waste. Our bin captures 2D, 3D, and thermal images of each waste disposal, and records environmental and mass data. Leveraging an onboard speech recognition model and a companion smartphone application, we enable users to annotate food waste images with a convenient, easy-to-use data collection framework. These compost bins will be deployed to 50 families across the state of Oregon in a forthcoming human subject study, where we aim to produce the first large, publicly available dataset of annotated food waste images. We leverage existing datasets of uneaten food item images to develop a computer vision model for food image recognition, towards the goal of producing an automatic food waste measurement system. Using these predictions, our app, CompostSense, presents users with an analytics dashboard summarizing their food waste trends, nudging home consumers to prevent food waste and adopt sustainable consumption habits.
Bio: I am a 4th year Ph.D. student in the AI/CS program. I’m advised by Dr. Patrick Donnelly, focusing on contrastive image-language models for food waste recognition. My postgrad goals are to apply my skills in industry, to combine AI/ML solutions with hardware to solve real-world problems for everyday people.