사각어묵 품질관리를 위한 비전기반 품질 검사 시스템의 개발 및 적용

Abstract

This research is on a computer vision-based automated quality inspection system for square fish cakes, aimed at enhancing consistency and productivity in seafood processing. The efficacy of the system was assessed in a simulated production environment, focusing on the detection and quantification of defects and foreign materials alongside of square fish cakes during processing of the food. Utilizing the computer vision model, YOLOv5 known for its real-time object detection capabilities, the study extends the application of these technologies within the scope of the seafood industry. The methodology involved constructing a comprehensive dataset constructed using images of square fish cakes contaminated with potential impurities such as plastic films and other residues found in seafood processing. The system underwent rigorous testing across various scenarios to replicate common production defects. Results demonstrated high accuracy, with precision and recall rates exceeding 95%, significantly enhancing product quality. The successful deployment of this system in a controlled environment showcases its potential to revolutionize seafood processing by reducing human error and increasing efficiency. This research advances the field of food quality control using computer vision while suggesting a scalable model for improving operational efficacy across the seafood industry.