IoT-Assisted Real-Time Detection of Post-Harvest Mango Diseases Using Image Processing, Machine Learning Validation, and an Automated Rotational Imaging System

Abstract

Mangoes are among the world's most valuable fruit crops, and postharvest disease can have a serious impact on quality, marketability, and storage life. Early detection of these diseases is necessary to diminish loss throughout a mango's lifespan – from the time it is harvested until it is consumed. This research study presents a real-time Internet of Things (IoT)-enabled detection system to find postharvest disease in mangoes utilizing image processing techniques and an automated rotating imaging mechanism. The proposed detection system is comprised of a rotating platform that rotates a camera around the circumference of the mango while capturing images of it from multiple perspectives. This camera setup gives the ability to visually inspect the entire surface of a mango throughout the entire 360° rotation of the ring. In addition to the image acquisition and rotation, the environmental conditions (temperature and humidity). The acquired images will be processed using image processing techniques including image acquisition, image preprocessing, image segmentation, feature extraction, and disease identification. This study focuses on three specific postharvest diseases of mangoes: Anthracnose, Stem-End Rot, and Transit Rot. Image processing is being used as the primary technique for detecting the presence of each disease, while machine learning algorithms are being used to confirm the classification results and provide performance metrics (e.g., accuracy about 91%) for the overall system. When a disease is identified as present via the above processes, the IoT module will generate a real-time notification.