本文已被：浏览 1462次 下载 448次
投稿时间：2015-07-20 修订日期：2016-03-22
投稿时间：2015-07-20 修订日期：2016-03-22
中文摘要: 现有的手绘草图识别方法严重依赖于费时费力的手工特征提取，而经典的深度学习模型主要是为彩色多纹理自然图像设计，用于识别手绘草图时效果不甚理想。本文提出了一种基于深度学习的手绘草图识别方法(Deep-Sketch) ，该算法根据手绘草图缺失颜色、纹理信息的特点，使用大尺寸的首层卷积核取代自然图像识别中常使用的小尺寸首层卷积核，获得更多的空间结构信息。利用训练浅层模型获得的模型参数来初始化深层模型对应层的模型参数，以加快收敛，减少训练时长。加入不改变特征大小的卷积层来加深网络深度等方法以减小错误率。实验结果表明，本文所提出的方法较之其它几种主流的手绘草图识别方法具有良好的正确率，对250类手绘草图识别正确率达到69.2%。
Abstract:In order to solve the existing problem of the sketch recognition heavily relying on the manual feature extraction which was very time-consuming, this paper proposed a method of sketch recognition based on deep learning, called Deep-Sketch. The classical deep learning models were mainly designed for natural color image recognition which failed on the sketch recognition. Deep-Sketch aimed to obtain more spatial structure information by using the large-size convolution kernel instead of the small-size convolution kernel in the first convolution layer. In addition, a shallow model was trained to obtain parameters which were used to initialize the corresponding layer parameters of the Deep-Sketch to reduce the model training time. Deep-Sketch was deepened with the convolution layers which kept the feature size to reduce the error rate. The results showed that the Deep-Sketch was superior to other state-of-the-art sketch recognition methods and achieved 69.2% accuracy on the sketch dataset including 250 classes
文章编号：201500716 中图分类号： 文献标志码：
基金项目:61472001基于车辆安全机制及评估方法研究；1408085MF122面向多样化图像检索的图像语义建模的研究
| 作者 | 单位 | |
| 赵鹏 | 安徽大学计算机科学与技术学院 | zhaopeng_ad@163.com |
| 王斐 | 安徽大学计算机科学与技术学院 | |
| 刘慧婷 | 安徽大学计算机科学与技术学院 | |
| 姚晟 | 安徽大学计算机科学与技术学院 |
| Author Name | Affiliation | |
| zhaopeng | zhaopeng_ad@163.com | |
作者简介:
引用文本：
赵鹏,王斐,刘慧婷,姚晟.基于深度学习的手绘草图识别[J].工程科学与技术,2016,48(3):94-99.
zhaopeng.Sketch Recognition Using Deep Learning[J].Advanced Engineering Sciences,2016,48(3):94-99.
引用文本：
赵鹏,王斐,刘慧婷,姚晟.基于深度学习的手绘草图识别[J].工程科学与技术,2016,48(3):94-99.
zhaopeng.Sketch Recognition Using Deep Learning[J].Advanced Engineering Sciences,2016,48(3):94-99.