Assessing the impact of epidermal rotting on postharvest mango quality: Utilizing multidimensional parameters and AI-based methods for damage classification

Mango is one of the most consumed fruits worldwide, yet it suffers significant postharvest losses due to epidermal rotting diseases associated with anthracnose and stem end rot. This study aimed to evaluate the impact of epidermal rotting (ER) on Tommy and Azucar mango genotypesacross multidimensional quality parameters during commercialization and postharvest. In addition, we propose an AI-based methods for classifying infected fruit using image analysis. Our study monitored ER in postharvest mangoes, assessed its impact on quality grading (Extra, First, Second, and Discard), and evaluated the associated economic impact.. In the laboratory, we quantified physical, chemical, and physiological quality variables in both infected and healthy mangoes. Furthermore, a consumer perception analysis was conducted using a Likert scale, complemented by natural language processing, to assess differences in consumer perceptions of healthy and infected fruits. We also trained and validated a set of three convolutional neural networks (CNN) to classify mangoes as healthy or infected based on images and a segmentation-based method to quantify the severity level. Our results show that ER significantly reduces the percentage of extra quality mangoes, by up to 50%, leading to fruit losses of 20-40% and a reduction in commercial value of 40-60%. Infected mangoes exhibited faster loss of firmness, increased respiration, and accelerated increases in total soluble solids and Brix levels. However, consumers did not detect the disease visually and reported no significant differences in consumption quality between healthy and infected fruits in the early stages of infection. CNN and segmentation demonstrated high accuracy for indirect detection and severity quantification of ER. This study highlights the substantial impact of ER on multidimensional quality parameters, underscoring the need for improved postharvest management practices.