AI-Powered Real-Time Rice Quality Inspection

One of the core innovations across my research is the development of intelligent computer vision systems capable of performing real-time rice quality inspection during the milling process. Traditional quality assessment relies heavily on manual sampling and post-processing inspection, which is slow, labour-intensive, and unable to prevent quality losses during production. My research replaces these conventional methods with deep learning models from the YOLO family that automatically detect, segment, measure, and classify rice grains while milling is in progress.

By combining advanced object detection, instance segmentation, and oriented bounding box techniques, the developed systems accurately identify whole, broken, and defective rice grains—even under challenging industrial conditions where grains overlap or appear at different orientations. This enables continuous, automated monitoring with high precision and provides a scalable solution for intelligent food manufacturing, significantly improving inspection speed, consistency, and product quality.