3rd Place Winner of Fall 2022
by Team DMV: Mohamed Gadelrab, Victoria Li, Darnell Khay
Abstract: Expired produce is an issue that continues to persist globally and cafeterias in Virginian schools are no exception. Our project aims to define several models that can successfully classify produce and identify whether each one is fresh or spoiled based on images. We made use of a multi-class model to classify the six different produce from the images dataset and another model for each produce which categorizes them through a fresh or spoiled class. Additionally, a model developed on Teachable Machine was also included for preliminary experimenting purposes. After proper data processing and training, our initial experiments have resulted in a functional Teachable Machine model that can classify tomatoes as spoiled or fresh. However, if given an image that does not contain a tomato, an image of a human for example, the model tries to determine whether the input image is a rotten or fresh tomato. This was concluded to be not of concern since the purpose of our research is to accurately classify produce and whether they are spoiled; not to recognize the difference from human faces and produce. Ultimately, the multi-class model and Teachable Machine model presented an outstanding accuracy which reveals a viable solution that can be used to combat the problem of mistaking rotten produce as fresh.
Comments