Abstract
The increasing rate of fresh produce spoiling while still being inside groceries are due to the produce already being overripe prior to being stocked. Improper harvesting of agricultural crops by automated laborers and overworked manual workers is the most probable cause for this issue. The goal of this thesis was to create an image recognition model to identify the current state of ripeness in a crop. Bananas were selected due to their visually distinct stages of ripeness. An online database of Bangladeshi bananas provided by Md Hasanul Ferdaus’s “BananaImageBD: A comprehensive banana image dataset for classification of banana varieties and detection of ripeness stages in Bangladesh”. This image recognition model uses OpenCV’s image processing library in order to determine the percentages of the different colors found on a banana peel. Ideally, this model will be extrapolated to support other crops with tighter margins in determining the quality of the produce.