计算机科学 ›› 2020, Vol. 47 ›› Issue (6): 114-120.doi: 10.11896/jsjkx.200300079
杨连平1, 孙玉波1, 张红良1, 李封2, 张祥德1
YANG Lian-ping1, SUN Yu-bo1, ZHANG Hong-liang1, LI Feng2, ZHANG Xiang-de1
摘要: 人体姿态估计尤其是多人姿态估计逐渐渗透到教育、体育等各个方面,精度高、轻量级的多人姿态估计更是当下的研究热点。自下而上的多人姿态估计方法的实时性较强,但是精度一般不高,网络结构也比较庞大。对于自下而上方法中最困难的一步——关键点关联问题,文中提出了一种轻量高效的姿态估计匹配网络。该网络在编码阶段将基础ResNet模块加以改进得到层结构,利用这些结构提取特征能够使得模型的参数量大幅减少;在解码阶段采用了特殊设计的反卷积结构,并在全网络添加了残差连接,这使得网络精度有很大的提升。整个算法能够将所有检测出来的关键点热图正确匹配到每一个人,得出最终的人体关键点估计。所提模型是一个轻便、高效的人体关键点匹配网络,它在COCO数据集地面真值上的mAP值高达89.7,而且参数只有8.01 M。这个结果相比目前最好的自下而上的人姿态估计方法在精度mAP值上提高了0.5,但是参数量仅为其1/10左右。所提网络利用COCO 2017和COCO 2014的地面真值分别进行了训练和验证,都取得了很高的精度,这证明其适合多种人体关键点热图的输入,并能够得到很好的效果。此外,文中针对网络模型的不同层结构设计了多种消融实验,最轻量的结构参数只有1.28兆,精度mAP值能够达到81.8。
中图分类号:
[1]CHEN Y,WANG Z,PENG Y,et al.Cascaded pyramid network for multi-person pose estimation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2018. [2]SUN K,XIAO B,LIU D,et al.Deep high-resolution representation learning for human pose estimation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2019. [3]CAO Z,SIMON T,WEI S E,et al.Realtime multi-person 2d pose estimation using part affinity fields [C] //Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2017. [4]KOCABAS M,KARAGOZ S,AKBAS E.Multiposenet:Fast multi-person pose estimation using pose residual network[C]//Proceedings of the European Conference on Computer Vision (ECCV).2018. [5]COCO Common Objects in Context[OL].http://cocodataset.org/. [6]XU H Y,YAO N M,PENG X L,et al.Multi-pose face image frontalization method based on codec network[J].Science in China:Information Science,2019,49(4):86-99. [7]HOWARD A G,ZHU M,CHEN B,et al.Mobilenets:Efficient convolutional neural networks for mobile vision applications[J].arXiv:1704.04861. [8]HE K,ZHANG X,REN S,et al.Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2016. [9]LONG J,SHELHAMER E,DARRELL T.Fully convolutional networks for semantic segmentation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2015:3431-3440. [10]TANG Z,PENG X,GENG S,et al.CU-net:coupled U-nets [J].arXiv:1808.06521. [11]ZEILER M D,FERGUS R.Visualizing and understanding convolutional networks[C]//European Conference on Computer Vision.Springer,Cham,2014:818-833. [12]ZEILER M D,KRISHNAN D,TAYLOR G W,et al.Deconvolutional networks[C]//2010 IEEE Computer Society Confe-rence on Computer Vision and Pattern Recognition.IEEE,2010. [13]XIAO B,WU H P,WEI Y C.Simple baselines for human pose estimation and tracking[C]//Proceedings of the European Conference on Computer Vision (ECCV).2018:466-481. [14]SZEGEDY C,VANHOUCKE V,IOFFE S,et al.Rethinking the inception architecture for computer vision[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2016. |
[1] | 邵延华, 李文峰, 张晓强, 楚红雨, 饶云波, 陈璐. 基于时空图卷积和注意力模型的航拍暴力行为识别 Aerial Violence Recognition Based on Spatial-Temporal Graph Convolutional Networks and Attention Model 计算机科学, 2022, 49(6): 254-261. https://doi.org/10.11896/jsjkx.210400272 |
[2] | 许华杰, 张晨强, 苏国韶. 基于深层卷积残差网络的航拍图建筑物精确分割方法 Accurate Segmentation Method of Aerial Photography Buildings Based on Deep Convolutional Residual Network 计算机科学, 2021, 48(8): 169-174. https://doi.org/10.11896/jsjkx.200500096 |
[3] | 杨紫淇, 蔡英, 张皓晨, 范艳芳. 基于负载均衡的VEC服务器联合计算任务卸载方案 Computational Task Offloading Scheme Based on Load Balance for Cooperative VEC Servers 计算机科学, 2021, 48(1): 81-88. https://doi.org/10.11896/jsjkx.200800220 |
[4] | 朱玲莹, 桑庆兵, 顾婷婷. 基于视差信息的无参考立体图像质量评价 No-reference Stereo Image Quality Assessment Based on Disparity Information 计算机科学, 2020, 47(9): 150-156. https://doi.org/10.11896/jsjkx.190700213 |
[5] | 冯晓月, 宋杰. 二维人体姿态估计研究进展 Research Advance on 2D Human Pose Estimation 计算机科学, 2020, 47(11): 128-136. https://doi.org/10.11896/jsjkx.200700061 |
[6] | 张晨,李志,朱红松,孙利民. 基于时空相关性的多签到数据匹配算法 MIMA:A Multi-identification Check-in Data Matching Algorithm Based on Spatial and Temporal Relations 计算机科学, 2018, 45(1): 188-195. https://doi.org/10.11896/j.issn.1002-137X.2018.01.033 |
[7] | 郑建彬,白雅贤,詹恩奇,汪阳. 基于改进SIFT匹配方法的货架乳制品识别 Improved SIFT Matching Method for Milk Beverage Recognition in Grocery 计算机科学, 2017, 44(9): 315-319. https://doi.org/10.11896/j.issn.1002-137X.2017.09.059 |
[8] | 李湘洋,赵杭生,赵小龙,张阳. 分布式多业务系统信道分配算法 Distributed Channel Allocation Algorithm for Multi-services Systems 计算机科学, 2016, 43(Z6): 272-275. https://doi.org/10.11896/j.issn.1002-137X.2016.6A.065 |
[9] | 李冬辉,张斌,费晓飞,刘洋. 基于多值属性分量的XACML策略匹配算法 Algorithm of Matching to XACML-Policy Based on Component of Multi-valued Attribute 计算机科学, 2014, 41(6): 104-107. https://doi.org/10.11896/j.issn.1002-137X.2014.06.021 |
[10] | 王浩,刘则芬,方宝富,陈金金. 基于约束树形图结构外观模型的人体姿态估计 Human Pose Estimation Based on Appearance Model for Constraint Tree Pictorial Structure 计算机科学, 2014, 41(3): 76-79. |
[11] | 丁山,宋丽晓. 一种改进的视网膜图像中微小动脉瘤的检测算法 Improved Method of Microaneurysm Detection Algorithm Based on Digital Fundus Images 计算机科学, 2014, 41(12): 269-274. https://doi.org/10.11896/j.issn.1002-137X.2014.12.058 |
[12] | 李 璋,杜慧敏,张丽果. 基于分布式存储的正则表达式匹配算法设计与实现 Fine-grained Parallel Multi-pattern Matching for Backbone Network NIDS 计算机科学, 2013, 40(3): 74-76. |
[13] | 祝晓东,郁松年. 运用视频技术的快速三维旋转分析与计算的研究 Research on Fast 3-D Rotation Calculation Based on Video Techniques 计算机科学, 2013, 40(2): 289-293. |
[14] | 禚 伟,金蓓弘,陈海彪,张利峰. 基于发布/订阅中间件的时空事件检测研究 Spatio-temporal Event Detection Using Pub/Sub Middleware 计算机科学, 2012, 39(10): 99-103. |
[15] | 王卓,冯晓宁,刘廷宝. DDM中基于历史信息排序的区域匹配算法 Region Matching Algorithm Based on Historical Information Sorting in DDM 计算机科学, 2011, 38(10): 202-204. |
|