基于稠密扩张卷积的图像语义分割模型  

Image Semantic Segmentation Model Based on Dense Dilation Convolution

作　　者：张富财 许建龙[1] 包晓安[1] ZHANG Fu-Cai;XU Jian-Long;BAO Xiao-An(School of Information Science and Technology,Zhejiang Sci-Tech University,Hangzhou 310018,China)

机构地区：[1]浙江理工大学信息学院,杭州310018

出　　处：《计算机系统应用》2022年第3期19-29,共11页Computer Systems & Applications

基　　金：浙江省重点研发计划(2020C03094)。

摘　　要：为解决图像语义分割任务中面对的分割场景的复杂性、分割对象的多样性及分割对象空间位置的差异性问题,提高语义分割模型的精度,提出基于稠密扩张卷积的双分支多层级语义分割网络(double branch and multistage network, DBMSNet).首先采用主干网络提取输入图像的4个不同分辨率的特征图(De1、De2、De3、De4),其次采用特征精炼(feature refine, FR)模块对De1和De3这两个特征图进行特征精炼处理,特征精炼处理之后的输出分支经过混合扩张卷积模块(mixed dilation module, MDM)编码空间位置特征, De4分支采用金字塔池化模块(pyramid pooling module, PPM)编码高级语义特征,最后将两个分支进行融合,输出分割结果.在数据集CelebAMask-HQ和Cityscapes中进行实验,分别得到mIoU精度为74.64%、78.29%.结果表明,本文方法的分割精度高于对比方法,且具有更少的参数量.Semantic segmentation is a very challenging task because of the complexity of parsing the scene, the diversity of segmented objects, and the differences in spatial positions of objects. To tackle this dilemma, this study proposes a novel architecture named double branch and multi-stage network(DBMSNet) based on dense dilation convolution.Firstly, four feature maps(De1, De2, De3, and De4) with different resolutions are extracted by the backbone network, and then the feature refinement maps of De1 and De3 are output through the feature refinement(FR) module. Secondly, the output branch is processed by the mixed dilation module(MDM) to extract rich spatial location features, while the De4 branch is processed by the pyramid pooling module(PPM) to extract multi-scale semantic information. Finally, the two branches are merged and the segmentation result is output. Comprehensive experiments are conducted on two public datasets of CelebAMask-HQ and Cityscapes, on which our model achieves mean intersection-over-union(mIoU) scores of 74.64% and 78.29%, respectively. The results show that the segmentation accuracy of this study is higher than that of the counterpart method, and this method has fewer parameters.

关 键 词：深度学习 图像语义分割 扩张卷积 稠密连接 多层级特征 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP18[自动化与计算机技术—计算机科学与技术]