深度学习与机器视觉在鱼类加工与品质监测中的研究进展

Research progress on deep learning and machine vision for fish processing and quality monitoring

在全球渔业产量不断增加和对鱼类品质保障需求提升的背景下,传统的鱼类加工和品质监测方法大多依赖人工操作,这不仅效率低下而且结果的一致性和准确性难以保证,逐渐无法满足现代需求。机器视觉和深度学习技术的结合,提供了一种高效、自动化的方法来提升鱼类加工与品质监测的准确性和效率。该综述概述了机器视觉系统和深度学习在鱼类加工中的应用,包括分类分拣、切割定位、重量估算等方面,并详细介绍了利用高光谱成像、近红外成像、比色传感器和传统成像等方法在品质监测中的最新研究进展,突出了深度学习在提升这些技术识别、分类精度和处理复杂图像数据能力方面的潜力。尽管机器学习技术在单一的加工问题中取得了成功,但面对复杂数据和环境变化时的适应性仍有限,这促使深度学习的相关研究日益受到重视。该文发现当前针对鱼类加工领域的深度学习研究还相对较少,且缺乏能够综合解决鱼类加工和品质监测多重任务的系统性研究。

As global fishery outputs grow and demands for fish quality assurance rise,traditional fish processing and quality monitoring methods are increasingly unable to meet modern requirements.The integration of machine vision and deep learning technologies presents an efficient and automated solution to enhance the accuracy and efficiency of fish processing and quality monitoring.This review outlines the applications of machine vision systems and deep learning in fish processing,including tasks such as sorting,cutting,and mass estimation.It delves into the latest research on quality monitoring using hyperspectral imaging,near-infrared imaging,colorimetric sensors,and traditional imaging,emphasizing the potential of deep learning to improve recognition,classification accuracy,and the processing of complex image data.Despite the success of machine learning in addressing specific processing issues,its adaptability to complex data and changing environmental conditions remains limited,underscoring the increasing importance of deep learning research.However,research in the fish processing domain that utilizes deep learning is still relatively sparse,with a notable absence of comprehensive systems capable of addressing multiple processing and monitoring challenges.This study found an urgent need for future research focused on developing integrated systems that could tackle a variety of tasks in fish processing and quality monitoring.Such systems promise not only to improve efficiency and reduce costs but also to ensure product quality through real-time surveillance.