Resource type
Thesis type
(Thesis) M.Sc.
Date created
2024-07-03
Authors/Contributors
Author (aut): Waseem, Muhammad Shahzaib
Abstract
Several studies show that a significant fraction of fresh fruits are discarded at the retail and consumer levels, wasting precious resources and polluting the environment. We propose a cost-effective, non-invasive solution (RipeTrack) that utilizes the sensing capabilities of smartphones and machine learning models to analyze fruits in different ripening stages. RipeTrack produces intuitive outputs, e.g., Unripe/Ripe/Expired and remaining lifetime (%), helping retailers and consumers to minimize food waste. We implement and deploy RipeTrack on various smartphones and demonstrate its accuracy using an extensive empirical study with multiple fruits. Our results show that RipeTrack can identify the ripeness of avocados and pears with an accuracy of 95% and 98%, respectively, and their remaining lifetimes with an accuracy of 93% and 95%. We also show that RipeTrack can be extended to new fruits using transfer learning, and it functions in realistic environments, e.g., homes and grocery stores, with diverse illuminations.
Document
Extent
58 pages.
Identifier
etd23151
Copyright statement
Copyright is held by the author(s).
Supervisor or Senior Supervisor
Thesis advisor (ths): Hefeeda, Mohamed
Language
English
Member of collection
Download file | Size |
---|---|
etd23151.pdf | 7.36 MB |