Revolutionizing Smart Agriculture: Enhancing Apple Quality with Machine Learning

Abstract

Agriculture 4.0 is a field that has spread widely around the world in this century, as it has undergone an exceptionally rapid evolution, especially when it comes to fruit recognition.Decisions about their quality are crucial to maximize profits and meet customer expectations. In the past, apples or even other fruits were based solely on visual assessments by experts, which led to errors. These old methods no longer consider the genetic evolution of apples, as they only consider their size, color, and skin imperfections. Digitizing this process saves energy and reduces costs and human error as well. Recent technological advances, which combine AI and CAO at the same time for fruit sorting, make it possible to achieve high levels of quality and meet the growing challenges of food safety on a global scale. This study proposes a machine learning-based multiclass model to improve the accuracy and efficiency of apple quality assessment. The model is trained on a large image dataset of three apple varieties: Gala, Fuji, and Golden Delicious (G.D). The model automatically classifies apples based on attributes such as color, shape, and imperfections, and evaluates their conformity. Experimental results demonstrate the effectiveness of this model, which achieves 97% accuracy in identifying apple varieties and assessing their quality. This approach significantly reduces inspection time and errors, optimizing operations in the production chain.

Abstract from the paper’s OpenAlex record; open copy licensed cc-by. Read at the source ↗

The text the model read, with the source record’s inline markup (italics, sub- and superscripts) rendered and its character entities decoded; quote offsets refer to the raw text.