[1]谭睿俊,赵志诚,谢新林.双层残差语义分割网络及交通场景应用[J].智能系统学报,2022,17(4):780-787.[doi:10.11992/tis.202106020]
TAN Ruijun,ZHAO Zhicheng,XIE Xinlin.Double-residual semantic segmentation network and traffic scenic application[J].CAAI Transactions on Intelligent Systems,2022,17(4):780-787.[doi:10.11992/tis.202106020]
点击复制
TAN Ruijun,ZHAO Zhicheng,XIE Xinlin.Double-residual semantic segmentation network and traffic scenic application[J].CAAI Transactions on Intelligent Systems,2022,17(4):780-787.[doi:10.11992/tis.202106020]
双层残差语义分割网络及交通场景应用
《智能系统学报》[ISSN 1673-4785/CN 23-1538/TP] 卷:
17
期数:
2022年第4期
页码:
780-787
栏目:
学术论文—人工智能基础
出版日期:
2022-07-05
- Title:
- Double-residual semantic segmentation network and traffic scenic application
- Keywords:
- double-residual network; detail feature extraction; jump feature fusion; up sampling; network branch training; semantic image segmentation; CamVid data set; Cityscapes data set
- 分类号:
- TP183
- 摘要:
- 针对图像语义分割过程中特征提取网络的深度问题以及下采样池化层降低特征图分辨率等问题，提出了一种基于双层残差网络特征提取的图像语义分割网络，称为DResnet。首先提出一种双层残差网络，对训练集各目标的细节进行特征提取，提高网络对部分细节目标的感知能力；其次在Layer1层开始跳跃特征融合，并持续以2倍反卷积方法进行上采样操作，融合底层特征与高层特征，降低部分细节信息丢失对分割精度的影响；最后使用网络分支训练法，先训练图像上各目标的大致轮廓特征，在此基础上再训练各目标的细节特征。结果表明：该网络的平均交并比较全卷积网络相比，在CamVid上由49.72%提升至59.44%，在Cityscapes上由44.35%提高到47.77%，该网络得到准确率更高、分割物体边缘更加完整的图像分割结果。
- Abstract:
- An image semantic segmentation network based on the double-residual network, named DRsenet, is proposed to address the depth problem of feature extraction network in semantic image segmentation and resolution reduction of feature maps by a down-sampling pooling layer. Firstly, a double-residual network is proposed to extract the detailed features of each target in the training set and improve the perception ability of the network for some detailed targets. Secondly, the jump feature fusion starts from Layer 1, and the up-sampling operation is continued by the 2×deconvolution method to fuse the low-level and the high-level features to reduce the impact of partial detail information loss on the segmentation accuracy. Finally, the network branch training method is adopted to train the outline features of each target on the image, followed by training the detailed features of each target. The results indicate that the network’s PMIOU improves from 49.72% to 59.44% on CamVid and from 44.35% to 47.77% on Cityscapes when compared to the full convolution network. The network can produce image segmentation results with higher accuracy and more complete edge segmentation.
- 参考文献/References:
-
[1] 朱磊, 滕奇志, 龚剑. 基于改进模糊C均值聚类和区域合并的矿物颗粒分割方法[J]. 科学技术与工程, 2020, 20(34): 14138–14145
ZHU Lei, TENG Qizhi, GONG Jian. Mineral particle segmentation algorithm based on improved fuzzy C-means and region merging[J]. Science technology and engineering, 2020, 20(34): 14138–14145
[2] 许林, 孟娜, 袁静, 等. 基于边缘检测和图像分割的超声诊断机器人控制系统设计[J]. 计算机测量与控制, 2020, 28(8): 125–129
XU Lin, MENG Na, YUAN Jing, et al. Design of ultrasonic diagnosis robot control system based on edge detection and image segmentation[J]. Computer measurement & control, 2020, 28(8): 125–129
[3] 陈飞. 改进的交互式Otsu红外图像分割算法[J]. 计算机测量与控制, 2020, 28(9): 248–251
CHEN Fei. An improved interactive otsu infrared image segmentation algorithm[J]. Computer measurement & control, 2020, 28(9): 248–251
[4] 杨金鑫, 杨辉华, 李灵巧, 等. 结合卷积神经网络和超像素聚类的细胞图像分割方法[J]. 计算机应用研究, 2018, 35(5): 1569–1572,1577
YANG Jinxin, YANG Huihua, LI Lingqiao, et al. Cell image segmentation method based on convolution neural network and super pixel clustering[J]. Application research of computers, 2018, 35(5): 1569–1572,1577
[5] KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks[J]. Communications of the acm, 2017, 60(6): 84–90.
[6] SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[C]//3rd International Conference on Learning Representations, ICLR 2015-Conference Track Proceedings. San Diego: [s. n. ], 2015.
[7] SZEGEDY C, LIU Wei, JIA Yangqing, et al. Going deeper with convolutions[C]//2015 IEEE Conference on Computer Vision and Pattern Recognition. Boston: IEEE, 2015: 1?9.
[8] TRONG V H, HYUN Y G, YOUNG K J, et al. Yielding multi-fold training strategy for image classification of imbalanced weeds[J]. Applied sciences, 2021, 11(8): 3331.
[9] MüLLER D, KRAMER F. MIScnn: a framework for medical image segmentation with convolutional neural networks and deep learning[EB/OL]. (2019?10?21)[2021?06?13].https://www.semanticscholar.org/paper/MIScnn%3A-A-Framework-for-Medical-Image-Segmentation-M%C3%BCller-Kramer.
[10] SHELHAMER E, LONG J, DARRELL T. Fully convolutional networks for semantic segmentation[C]//IEEE transactions on pattern analysis and machine intelligence. New York: IEEE, 2015 : 640?651.
[11] EVERINGHAM M, ESLAMI S M A, GOOL L, et al. The pascal visual object classes challenge: a retrospective[J]. International journal of computer vision, 2015, 111(1): 98–136.
[12] SILBERMAN N, HOIEM D, KOHLI P, et al. Indoor segmentation and support inference from RGBD images[C]//European Conference on Computer Vision. Berlin: Springer, 2012: 746?760.
[13] RONNEBERGER O, FISCHER P, BROX T. U-net: convolutional networks for biomedical image segmentation[M]//Lecture Notes in Computer Science. Cham: Springer International Publishing, 2015: 234?241.
[14] XIAO Zhitao, LIU Bowen, GENG Lei, et al. Segmentation of lung nodules using improved 3D-UNet neural network[J]. Symmetry, 2020, 12(11): 1787.
[15] GUAN Haixing, LI Hongliang, LI Rongqiang, et al. Face detection of innovation base based on faster RCNN[M]//2021 International Conference on Applications and Techniques in Cyber Intelligence. Cham: Springer International Publishing, 2021: 158?165.
[16] HE Kaiming, ZHANG Xiangyu, REN Shaoqing, et al. Delving deep into rectifiers: surpassing human-level performance on ImageNet classification[C]//2015 IEEE International Conference on Computer Vision. Santiago: IEEE, 2015: 1026?1034.
[17] MANDAL R, AZAM B, VERMA B, et al. Deep learning model with GA-based visual feature selection and context integration[C]//2021 IEEE Congress on Evolutionary Computation. Kraków: IEEE, 2021: 288?295.
[18] YANG Tao, WU Yan, ZHAO Junqiao, et al. Semantic segmentation via highly fused convolutional network with multiple soft cost functions[J]. Cognitive systems research, 2019, 53: 20–30.
[19] JIA Fan, LIU Jun, TAI Xuecheng. A regularized convolutional neural network for semantic image segmentation[J]. Analysis and applications, 2021, 19(1): 147–165.
[20] CORDTS M, OMRAN M, RAMOS S, et al. The cityscapes dataset for semantic urban scene understanding[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas: IEEE, 2016: 3213?3223.
[21] KAZEROUNI I A, DOOLY G, TOAL D. Ghost-UNet: an asymmetric encoder-decoder architecture for semantic segmentation from scratch[J]. IEEE access, 2021, 9: 97457–97465.
备注/Memo
收稿日期:2021-06-13。
基金项目:山西省自然科学基金青年基金项目（201901D211304）.
作者简介:谭睿俊，硕士研究生，主要研究方向为机器视觉与智能信息处理。曾获国家奖学金，“华为杯”第十七届中国研究生数学建模竞赛三等奖，“兆易创新杯”第十六届中国研究生电子设计竞赛三等奖，全国大学生智能汽车竞赛–讯飞智慧餐厅组别线上国家三等奖，第十六届全国大学生智能汽车竞赛分赛区讯飞智慧餐厅组比赛国家二等奖;赵志诚，教授，主要研究方向为先进控制技术。承担并完成国家级、省部级以及企业委托项目40余项，获省部级科技奖励2项，发表学术论文80余篇，出版专著1部，授权发明专利4项;谢新林，讲师，主要研究方向为机器视觉、深度学习。主持省级科研项目2项、横向课题1项，发表学术论文10余篇、申请和授权专利10余项
通讯作者:赵志诚. E-mai：zhzhich@126.com
更新日期/Last Update:
1900-01-01