【摘要】 随着大数据时代的到来,作为类脑计算领域的一个重要研究成果,卷积神经网络(convolutnioal neural networks,CNNs)已广泛应用于多个领域。与传统机器学习相比,卷积神经网络拥有更复杂的网络结构和更多隐藏层,有更强的特征学习和特征表达的能力,已被较好地运用处理多个大规模分类识别任务。目前纺织服装行业大量丰富的图像数据正好迎合了它的应用,已经有许多纺织服装图像领域的研究运用卷积神经网络技术,并取得了较好的效果。梳理了卷积神经网络运用在图像分类和目标检测两方面的主要经典网络结构,并分别介绍了这些网络为更好地应用于纺织服装领域而进行的改良与创新,最后结合现阶段发展给出未来可以运用的理论方向。更多还原

【Abstract】 With the advent of big data era,as an important research achievement in brain-like computing,convolutional neural networks(CNN) has been widely applied in many fields.Compared with traditional machine learning,convolutional neural network has more complex network structure and more hidden layers,and owns stronger feature learning ability and feature expression ability.It has been used to deal with many large-scale classification and recognition tasks.At present,a large number of rich image data in the textile and clothing industry just meet its application,and many researches in the field of textile and garment image have adopted convolutional neural network technology and achieved good results.This paper summarized the main classical network structures of convolutional neural network used in image classification and target detection,and introduced the improvement and innovation of these networks for better application in the field of textile and clothing.Finally,the future theoretical direction was provided based on the current development.更多还原