Intelligent grading system for mangosteen based on faster-FRNet: Enhancing accuracy and efficiency in post-harvest quality control

Abstract

This study addresses the inefficiencies and errors associated with traditional manual grading of mangosteen by introducing an advanced automated grading system. The system combines specialized hardware and a novel machine learning approach using the Faster-FRNet model, an enhanced version of Faster R-CNN integrated with ResNet50 and Feature Pyramid Network (FPN) for improved multi-scale feature extraction. The primary purpose is to develop a high-accuracy, scalable grading solution. Experimental results demonstrate a grading accuracy of 98.75 %, with a mean Average Precision (mAP) of 0.68, surpassing Faster R-CNN with ResNet50 (0.51 mAP) and VGG16 (0.45 mAP). The system not only improves grading accuracy and speed but also reduces computational complexity, making it suitable for large-scale agricultural applications. To summarize, this research advances the field of smart agricultural systems by presenting a versatile approach that improves post-harvest management, boosts economic profitability, and elevates customer satisfaction. The proposed solution holds promise for widespread adoption across diverse agricultural commodities.