A Real-Time Automated System for Shrimp Quality Assessment Using YOLOv10

A real-time system for shrimp quality evaluation is developed to enhance quality control in the seafood industry. The study is designed to reduce manual errors and improve operational efficiency by automatically classifying shrimp into nine distinct categories that distinguish between defective and high-quality specimens. The methodology is based on the YOLOv10 object detection algorithm, which is augmented by advanced image preprocessing techniques-such as data augmentation and precise bounding box annotation-to increase model robustness. Integration with Streamlit and FastAPI is performed to enable scalable, user-friendly deployment in industrial environments. Major findings reveal that the system achieves a mean Average Precision (mAP) at an IoU threshold of 0.5 (mAP@50) of 99.15% and mean Average Precision at IoU thresholds ranging from 0.5 to 0.95 (mAP@50-95) of 92.9%, demonstrating significant improvements in detection accuracy, speed, and computational efficiency compared to earlier versions. These results suggest that the proposed system can substantially reduce human error, optimize throughput, and support rigorous quality assurance processes, thereby promoting competitiveness and customer satisfaction in the global seafood market.