| Abstract: |
The quality assessment of juicy fruits such as oranges, mangoes, watermelons, and pomegranates is traditionally performed through manual, subjective, and destructive testing methods that are time-consuming and prone to human error. This paper proposes an Internet of Things (IoT)-based framework integrated with Machine Learning (ML) algorithms for non-destructive, real-time quality grading of juicy fruits based on parameters such as sugar content (Brix), firmness, moisture, color intensity, and pH level [1]-[3]. A network of IoT sensors including near-infrared (NIR) spectroscopy modules, load cells, humidity sensors, and RGB color sensors was deployed to collect fruit-specific data, which was transmitted via a microcontroller (ESP32/NodeMCU) to a cloud platform for storage and analysis [4]. Five machine learning classifiers Support Vector Machine (SVM), Random Forest (RF), K-Nearest Neighbor (KNN), Decision Tree (DT), and Convolutional Neural Network (CNN) were trained on the collected dataset to classify fruit quality into categories such as Premium, Standard, and Reject. Experimental results demonstrate that the Random Forest and CNN models achieved the highest classification accuracy of 96.4% and 97.1% respectively, outperforming traditional statistical grading methods. The correlation between sensor-derived features and actual laboratory-tested quality parameters was found to be statistically significant (r > 0.90). These findings validate the effectiveness of combining IoT sensing with ML-based analytics for scalable, accurate, and real-time fruit quality monitoring, offering practical value for the agricultural supply chain, post-harvest management, and smart farming ecosystems. |
