摘要

【目的】 将人工智能方法引入数字人文领域中,探讨如何解决中华传统刺绣图像分类背景下刺绣数据集较小、图像特征表示不足以及识别准确率不高等问题,为非物质文化遗产数字保护智能化提供方法支撑。 【方法】 将深度学习技术运用到刺绣图像上,利用图像处理技术提取其相应的特征,采用迁移学习的方法,对Xception模型进行微调改进,进而提出一种基于Xception-TD的中华传统刺绣分类模型,并探讨全连接层的数量与维度以及dropout取值对模型性能的影响。 【结果】 实验结果表明,针对中华传统刺绣分类的问题,通过微调的方法,发现提高全连接层数量以及增大全连接层维度可以得到更好的刺绣图像特征表示并产生更好的效果。基于Xception-TD中华传统刺绣模型准确率达到0.968 63,均优于基准模型。在进一步刺绣多分类的问题上,准确率也均优于基准模型。 【局限】 本文数据集仅来源于百度图片与少量人工标记,数据来源不够丰富。 【结论】 基于迁移学习,并结合微调能够有效提升刺绣分类的准确率。

Abstract ：

[Objective] This paper introduces artificial intelligence methods to the field of digital humanities, aiming to address the issues of small data sets, insufficient image feature representation, and low recognition accuracy facing traditional Chinese embroidery image classification. It also tries to prvovide methodology support to the digitalization of intangible cultural heritage protection. [Methods] We utilized deep learning techniques to analyze the embroidery images, and extracted their features. Then, we fine-tuned the Xception model with the migration learning approach, and constructed a Xception-TD method to classify traditional Chinese embroidery. Finally, we explored the impacts of the number and dimensions of fully connected layers, as well as the value of dropouts on the model’s performance. [Results] We found that increasing the number and dimensions of fully connected layers improved the embroidery image feature representation. The accuracy rate of our new model reached 0.96863, which was better than the benchmark model. In multi-classification tasks, the model’s accuracy was also better than that of the benchmark ones. [Limitations] The experimental data set was only constructed with Baidu images, which had small amount of manual taggings. [Conclusions] The proposed model based on transfer learning could improve the accuracy of embroidery classification.

Key words ： Digital Humanities Computer Vision Transfer Learning Xception 周泽聿, 王昊, 张小琴, 范涛, 任秋彤. 基于Xception-TD的中华传统刺绣分类模型构建 ^* [J]. 数据分析与知识发现, 2022, 6(2/3): 338-347.
Zhou Zeyu, Wang Hao, Zhang Xiaoqin, Tao Fao, Ren Qiutong. Classification Model for Chinese Traditional Embroidery Based on Xception-TD. Data Analysis and Knowledge Discovery, 2022, 6(2/3): 338-347.

序号	模型	模型介绍	准确率
1	未微调的Xception	使用ImageNet预训练好的Xception直接作为特征提取的模型参数,通过Softmax层对目标数据集中华传统刺绣进行分类。	0.951 07
2	未微调的VGG-19	VGG网络体系结构最初是由Simonyan和Zisserman提出的,其中VGG19主要架构是5个卷积层块与三个全连接层 [ 41 ] 。	0.942 28
3	未微调的Xception-SVM	使用ImageNet预训练好的Xception直接作为特征提取的模型参数,通过支持向量机SVM对目标数据集中华传统刺绣进行分类。	0.940 32

Lowe D G. Distinctive Image Features from Scale-Invariant Keypoints[J]. International Journal of Computer Vision, 2004, 60(2):91-110. doi: 10.1023/B:VISI.0000029664.99615.94 Tajeripour F, Saberi M, Fekri-Ershad S. Developing a Novel Approach for Content Based Image Retrieval Using Modified Local Binary Patterns and Morphological Transform[J]. International Arab Journal of Information Technology, 2015, 12(6):574-581. Yeh C H, Lin M H, Chang P C, et al. Enhanced Visual Attention-Guided Deep Neural Networks for Image Classification[J]. IEEE Access, 2020, 8:163447-163457. doi: 10.1109/ACCESS.2020.3021729 Simonyan K, Zisserman A. Very Deep Convolutional Networks for Large-Scale Image Recognition[OL]. arXiv Preprint,arXiv: 1409.1556. He K M, Zhang X Y, Ren S Q, et al. Deep Residual Learning for Image Recognition[C]// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. IEEE, 2016: 770-778. Chollet F. Xception: Deep Learning with Depthwise Separable Convolutions[C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition. IEEE, 2017: 1800-1807. 牛鑫鑫, 孙阿猛, 王钎沣, 等. 基于深度学习的遥感图像分类研究[J]. 激光杂志, 2021, 42(5):10-14. ( Niu Xinxin, Sun Ameng, Wang Qianfeng, et al. Study on Remote Sensing Image Classification Based on Deep Learning[J]. Laser Journal, 2021, 42(5):10-14.) Alzubaidi L, Zhang J L, Humaidi A J, et al. Review of Deep Learning: Concepts, CNN Architectures, Challenges, Applications, Future Directions[J]. Journal of Big Data, 2021, 8(1):53. doi: 10.1186/s40537-021-00444-8 pmid: 33816053 李赵旭, 宋涛, 葛梦飞, 等. 基于改进Inception模型的乳腺癌病理学图像分类[J]. 激光与光电子学进展, 2021, 58(8):0817001. ( Li Zhaoxu, Song Tao, Ge Mengfei, et al. Breast Cancer Classification from Histopathological Images Based on Improved Inception Model[J]. Laser & Optoelectronics Progress, 2021, 58(8):0817001.) 高淑萍, 赵清源, 齐小刚, 等. 改进MobileNet的图像分类方法研究[J]. 智能系统学报, 2021, 16(1):11-20. ( Gao Shuping, Zhao Qingyuan, Qi Xiaogang, et al. Research on the Improved Image Classification Method of MobileNet[J]. CAAI Transactions on Intelligent Systems, 2021, 16(1):11-20.) 谢豪, 毛进, 李纲. 基于多层语义融合的图文信息情感分类研究[J]. 数据分析与知识发现, 2021, 5(6):103-114. ( Xie Hao, Mao Jin, Li Gang. Sentiment Classification of Image-Text Information with Multi-Layer Semantic Fusion[J]. Data Analysis and Knowledge Discovery, 2021, 5(6):103-114.) 范涛, 吴鹏, 王昊, 等. 基于多模态联合注意力机制的网民情感分析研究[J]. 情报学报, 2021, 40(6):656-665. ( Fan Tao, Wu Peng, Wang Hao, et al. Sentiment Analysis of Online Users Based on Multimodal Co-Attention[J]. Journal of the China Society for Scientific and Technical Information, 2021, 40(6):656-665.) 乔思波, 庞善臣, 王敏, 等. 基于残差混合注意力机制的脑部CT图像分类卷积神经网络模型[J]. 电子学报, 2021, 49(5):984-991. doi: 10.12263/DZXB.20200881 ( Qiao Sibo, Pang Shanchen, Wang Min, et al. A Convolutional Neural Network for Brain CT Image Classification Based on Residual Hybrid Attention Mechanism[J]. Acta Electronica Sinica, 2021, 49(5):984-991.) doi: 10.12263/DZXB.20200881 许骞艺, 秦贵和, 孙铭会, 等. 基于改进的ResNeSt驾驶员头部状态分类算法[J]. 吉林大学学报(工学版), 2021, 51(2):704-711. ( Xu Qianyi, Qin Guihe, Sun Minghui, et al. Classification of Drivers' Head Status Based on Improved ResNeSt[J]. Journal of Jilin University (Engineering and Technology Edition), 2021, 51(2):704-711.) 陈巧红, 陈翊, 李文书, 等. 多尺度SE-Xception服装图像分类[J]. 浙江大学学报(工学版), 2020, 54(9):1727-1735. ( Chen Qiaohong, Chen Yi, Li Wenshu, et al. Clothing Image Classification Based on Multi-Scale SE-Xception[J]. Journal of Zhejiang University(Engineering Science), 2020, 54(9):1727-1735.) Zhe Z, Wang Q H, Xing Y D. Research on Big Data Analysis Technology of Chinese Traditional Culture Yue Embroidery Color Network[J]. Journal of Physics: Conference Series, 2019, 1345(2):022021. doi: 10.1088/1742-6596/1345/2/022021 张效娟. 非物质文化遗产的数字化保护与开发: 以青海刺绣艺术为例[J]. 青海社会科学, 2018(3):201-204. ( Zhang Xiaojuan. Digital Protection and Development of Intangible Cultural Heritage: The Case of Qinghai Embroidery Art[J]. Qinghai Social Sciences, 2018(3):201-204.) 刘净净, 郭飞, 刘玉. 刺绣图片的计算机智能识别[J]. 电脑知识与技术, 2012, 8(35):8483-8486. ( Liu Jingjing, Guo Fei, Liu Yu. The Computer Intelligent Recognition of Embroidery Pictures[J]. Computer Knowledge and Technology, 2012, 8(35):8483-8486.) 龚伟伟. 基于卷积神经网络的刺绣图像检索研究[D]. 西宁: 青海师范大学, 2020. ( Gong Weiwei. Research on Embroidery Image Retrieval Based on Convolutional Neural Network[D]. Xining: Qinghai Normal University, 2020.) 杨蕾, 胡慧, 周军. 刺绣针法图样特征点提取及匹配方法研究[J]. 计算机应用研究, 2021, 38(7):2231-2234. ( Yang Lei, Hu Hui, Zhou Jun. Study on Feature Points Extraction and Matching of Stitch Pattern[J]. Application Research of Computers, 2021, 38(7):2231-2234.) 赵凯琳, 靳小龙, 王元卓. 小样本学习研究综述[J]. 软件学报, 2021, 32(2):349-369. ( Zhao Kailin, Jin Xiaolong, Wang Yuanzhuo. Survey on Few-Shot Learning[J]. Journal of Software, 2021, 32(2):349-369.) 陈虹丽, 刘凌风, 李浩凯, 等. 基于迁移学习的指关节纹识别算法实验设计[J]. 实验技术与管理, 2021, 38(6):81-84. ( Chen Hongli, Liu Lingfeng, Li Haokai, et al. Experimental Design of Knuckle Pattern Recognition Algorithm Based on Transfer Learning[J]. Experimental Technology and Management, 2021, 38(6):81-84.) 张德军, 周学成, 杨旭东. 基于图像处理和深度迁移学习的芒果果实病状识别[J]. 华南农业大学学报, 2021, 42(4):113-124. ( Zhang Dejun, Zhou Xuecheng, Yang Xudong. Recognition of Mango Fruit Diseases Based on Image Processing and Deep Transfer Learning[J]. Journal of South China Agricultural University, 2021, 42(4):113-124.) 陈德刚, 艾孜尔古丽, 尹鹏博, 等. 基于改进Xception迁移学习的野生菌种类识别研究[J]. 激光与光电子学进展, 2021, 58(8):0810023. ( Chen Degang, Azragul, Yin Pengbo, et al. Research on Identification of Wild Mushroom Species Based on Improved Xception Transfer Learning[J]. Laser & Optoelectronics Progress, 2021, 58(8):0810023.) 蒋雨肖, 丁晟春, 吴鹏. 基于BiLSTM-VGG16的多模态信息特征分类研究[J]. 情报理论与实践, 2021, 44(11):180-186. ( Jiang Yuxiao, Ding Shengchun, Wu Peng. A Study on the Classification of Features of Multi-Modal Information Based on BiLSTM-VGG16[J]. Information Studies: Theory & Application, 2021, 44(11):180-186.) 李昆仑, 王怡辉, 陈栋, 等. 结合注意力与双线性网络的细粒度图像分类[J]. 小型微型计算机系统, 2021, 42(5):1071-1076. ( Li Kunlun, Wang Yihui, Chen Dong, et al. Combines Attention with Bilinear Networks for Fine-Grained Image Classification[J]. Journal of Chinese Computer Systems, 2021, 42(5):1071-1076.) 樊湘鹏, 许燕, 周建平, 等. 基于迁移学习和改进CNN的葡萄叶部病害检测系统[J]. 农业工程学报, 2021, 37(6):151-159. ( Fan Xiangpeng, Xu Yan, Zhou Jianping, et al. Detection System for Grape Leaf Diseases Based on Transfer Learning and Updated CNN[J]. Transactions of the Chinese Society of Agricultural Engineering, 2021, 37(6):151-159.) Carvalho T, de Rezende E R S, Alves M T P, et al. Exposing Computer Generated Images by Eye’s Region Classification via Transfer Learning of VGG19 CNN[C]// Proceedings of 2017 16th IEEE International Conference on Machine Learning and Applications. IEEE, 2017: 866-870. Srivastava N, Hinton G, Krizhevsky A, et al. Dropout: A Simple Way to Prevent Neural Networks from Overfitting[J]. Journal of Machine Learning Research, 2014, 15(1):1929-1958. 张卫, 王昊, 陈玥彤, 范涛, 邓三鸿. 融合迁移学习与文本增强的中文成语隐喻知识识别与关联研究* [J]. 数据分析与知识发现, 2022, 6(2/3): 167-183. 陆泉, 何超, 陈静, 田敏, 刘婷. 基于两阶段迁移学习的多标签分类模型研究* [J]. 数据分析与知识发现, 2021, 5(7): 91-100. 张琪,江川,纪有书,冯敏萱,李斌,许超,刘浏. 面向多领域先秦典籍的分词词性一体化自动标注模型构建 ^* [J]. 数据分析与知识发现, 2021, 5(3): 2-11. 王倩,王东波,李斌,许超. 面向海量典籍文本的深度学习自动断句与标点平台构建研究 ^* [J]. 数据分析与知识发现, 2021, 5(3): 25-34. 纪有书, 王东波, 黄水清. 基于词对齐的古汉语同义词自动抽取研究 ^* ——以前四史典籍为例 [J]. 数据分析与知识发现, 2021, 5(11): 135-144. 赵宇翔,练靖雯. 数字人文视域下文化遗产众包研究综述* [J]. 数据分析与知识发现, 2021, 5(1): 36-55. 梁继文,江川,王东波. 基于多特征融合的先秦典籍汉英句子对齐研究 ^* [J]. 数据分析与知识发现, 2020, 4(9): 123-132. 徐晨飞, 叶海影, 包平. 基于深度学习的方志物产资料实体自动识别模型构建研究* [J]. 数据分析与知识发现, 2020, 4(8): 86-97. 赵平,孙连英,涂帅,卞建玲,万莹. 改进的知识迁移景点实体识别算法研究及应用 ^* [J]. 数据分析与知识发现, 2020, 4(5): 118-126. 刘彤,倪维健,孙宇健,曾庆田. 基于深度迁移学习的业务流程实例剩余执行时间预测方法 ^* [J]. 数据分析与知识发现, 2020, 4(2/3): 134-142. 向菲,谢耀谈. 基于混合采样与迁移学习的患者评论识别模型* [J]. 数据分析与知识发现, 2020, 4(2/3): 39-47. 刘浏,秦天允,王东波. 非物质文化遗产传统音乐术语自动抽取 ^* [J]. 数据分析与知识发现, 2020, 4(12): 68-75. 王树义,刘赛,马峥. 基于深度迁移学习的微博图像隐私分类研究 ^* [J]. 数据分析与知识发现, 2020, 4(10): 80-92. 地址：北京市海淀区中关村北四环西路33号 邮编：100190
电话/传真：(010)82626611-6626，82624938