摘要
纺织业中棉麻混纺制品具有透气性、透汗性强等优点,但其中的棉麻纤维配比会影响实际制品的质量,而为了检测纤维配比需要对检测人员经过专业训练,并通过人眼观察来判断检测棉麻混纺纱中棉麻纤维比例。这存在着人力资源与时间资源损耗大,且主观影响因素过强等缺点,为了解决这个问题,本文提出利用神经网络来代替人工自动检测棉麻混纺纱中的棉麻比例,在经过试验后,利用YOLOv3网络并经过一定改良,对于测试集数据的最优平均精度均值(mAP)达到0.973,且测得棉麻混纺比平均值符合实际样本棉麻混纺比数值。由此可以证明,利用神经网络对棉麻混纺纱进行自动的比例检测是成立的。
Cotton flax blended products in the textile industry have the advantages of good air permeability and sweat permeability,but the cotton flax fiber ratio would affect the quality of actual products. In order to detect the fiber ratio, the inspectors need to be professionally trained and detect the cotton flax fiber ratio in cotton flax blended yarn through subjective judgment of human eyes,which has a large loss of human resources and time resources, and the subjective influencing factors are too strong. In order to solve these problems, it is proposed to use neural network instead of manual automatic detection of cotton flax ratio in cotton flax blended yarn. After the test, the mean average precision(MAP) of the network for the test set data is 0.973, which obtained by using improved yolov3 network, and the average value of cotton linen blending ratio measured is consistent with the value of cotton linen blending ratio of actual samples. Therefore, it can be proved that it is feasible to use neural network detection automatically.
作者
刘瀚旗
邓中民
李相朋
LIU Han-qi;DENG Zhong-min;LI Xiang-peng(Engineering Research Center of Hubei Province for Clothing Information,Wuhan Hubei 430200 China;State Key Laboratory of New Textile Materials and Advanced Processing Technology,Wuhan Textile University,Wuhan Hubei 430200,China;School of Computer Science and Artificial Intelligence,Wuhan Textile University,Wuhan Hubei 430200,China)
出处
《武汉纺织大学学报》
2021年第6期3-8,共6页
Journal of Wuhan Textile University
关键词
棉麻检测
神经网络
深度学习
目标检测
cotton and flax detection
neural network
deep learning
object detection
