改进Mask RCNN算法及其在行人实例分割中的应用

doi:10.19678/j.issn.1000-3428.0058058

摘要/Abstract

摘要： Mask RCNN算法在特征提取过程中存在语义信息丢失的问题，而自然场景中的行人具有姿态不同、遮挡和背景复杂等特点，导致算法应用于行人实例分割时检测准确性较差。对此，提出一种改进的Mask RCNN算法。在Mask RCNN网络的Mask分支中增加串联特征金字塔网络（CFPN）模块，对网络生成的多层特征进行融合，充分利用不同特征层的语义信息。在此基础上，执行RoI Align操作生成行人掩膜。仿照COCO数据集，从生活场景中拍摄1 000张图片，自建一个新的行人数据集。基于该数据集的实验结果表明，改进算法较原算法具有更高的检测精确率。

关键词: 行人实例分割, Mask RCNN算法, 特征融合, 目标检测, 串联特征金字塔网络

Abstract: In the process of feature extraction, Mask RCNN algorithm will lose semantic information.However, pedestrians in natural scenes have different posture, occlusion and complex background, which lead to poor detection accuracy in pedestrian segmentation.To solve this problem, an improved Mask RCNN algorithm is proposed.In the Mask branch of the Mask RCNN network, the Concatenated Feature Pyramid Network(CFPN) is added to fuse the multi-layer features generated by the network, so as to make full use of the semantic information of different feature layers.On this basis, the RoI Align operation is performed to generate a pedestrian mask.Following the COCO data set, 1 000 pictures are taken from life scenes, and a new pedestrian data set is built.Experimental results based on the data set show that the improved algorithm has higher detection accuracy than the original algorithm.

Key words: pedestrian instance segmentation, Mask RCNN algorithm, feature fusion, object detection, Concatenated Feature Pyramid Network(CFPN)

中图分类号:

TP391

音松, 陈雪云, 贝学宇. 改进Mask RCNN算法及其在行人实例分割中的应用[J]. 计算机工程, 2021, 47(6): 271-276,283.

YIN Song, CHEN Xueyun, BEI Xueyu. Improved Mask RCNN Algorithm and Its Application in Pedestrian Instance Segmentation[J]. Computer Engineering, 2021, 47(6): 271-276,283.

https://www.ecice06.com/CN/Y2021/V47/I6/271

图/表 11

20210618184829

20210618184833

20210618184836

20210618184841

20210618184844

20210618184847

20210618184850

20210618184854

20210618184857

20210618184900

20210618184904

参考文献

[1] HU Jianqiu,XING Xianglei,JIANG Pan,et al.Pedestrian segmentation based on Mask R-CNN[J].Command Control and Simulation,2020(5):42-46.(in Chinese)胡剑秋,邢向磊,蒋攀,等.基于Mask R-CNN的行人分割[J].指挥控制与仿真,2020(5):42-46.
[2] WEN Yaole,LI Linyan,SHANG Xinru,et al.An instance segmentation method based on improved Mask RCNN feature fusion[J].Computer Applications and Software,2019,36(10):130-133.(in Chinese)温尧乐,李林燕,尚欣茹,等.一种改进的Mask RCNN特征融合实例分割方法[J].计算机应用与软件,2019,36(10):130-133.
[3] YU Bo,MA Shuhao,LI Hongyan,et al.Real-time pedestrian detection for far-infrared vehicle images and adaptive instance segmentation[J].Laser & Optoelectronics Progress,2020,57(2):293-303.(in Chinese)于博,马书浩,李红艳,等.远红外车载图像实时行人检测与自适应实例分割[J].激光与光电子学进展,2020,57(2):293-303.
[4] ZHANG Yugui,SHEN Liuqing,HU Haimiao.Extraction of foreground area of pedestrian objects under thermal infrared video surveillance[J].Journal of Beijing University of Aeronautics and Astronautics,2020,46(9):1721-1729.(in Chinese)张玉贵,沈柳青,胡海苗.热红外视频监控下行人目标前景区域提取[J].北京航空航天大学学报,2020,46(9):1721-1729.
[5] LUO Yuan,WANG Boyu,CHEN Xu.Research progresses of target detection technology based on deep learning[J].Semiconductor Optoelectronics,2020,41(1):1-10.(in Chinese)罗元,王薄宇,陈旭.基于深度学习的目标检测技术的研究综述[J].半导体光电,2020,41(1):1-10.
[6] GIRSHICK R,DONAHUE J,DARRELL T,et al.Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of 2014 IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C,USA:IEEE Press,2014:1-5.
[7] GIRSHICK R.Fast R-CNN[C]//Proceedings of 2015 IEEE International Conference on Computer Vision.Washington D.C,USA:IEEE Press,2015:1-5.
[8] SIMONYAN K,ZISSERMAN A.Very deep convolutional networks for large-scale image recognition[EB/OL].(2015-04-10)[2020-04-22].https://arxiv.org/pdf/1409.1556.pdf.
[9] REN S,HE K,GIRSHICK R,et al.Faster R-CNN:towards real-time object detection with region proposal networks[C]//Proceedings of the 28th International Conference on Neural Information Processing Systems.New York,USA:ACM Press,2015:91-99.
[10] JOSEPH R,SANTOSH D.YOLO:real-time object detection[EB/OL].(2016-11-03)[2020-04-22].http://pjreddie.com/darknet.
[11] REDMON J,FARHADI A.YOLO9000:better,faster,stronger[C]//Proceedings of 2017 IEEE Conference on Com-puter Vision and Pattern Recognition.Washington D.C.,USA:IEEE Press,2017:6517-6525.
[12] REDMON J,FARHADI A.YOLOv3:an incremental improvement[C]//Proceedings of 2018 IEEE Con-ference on Computer Vision and Pattern Recognition.Washington D.C.,USA:IEEE Press,2018:1-5.
[13] SZEGEDY C,LIU W,JIA Y,et al.Going deeper with convolutions[C]//Proceedings of 2015 IEEE Conference on Computer Vision and Pattern.Washington D.C.,USA:IEEE Press,2015:1-5.
[14] CHEN L C,PAPANDREOU G,KOKKINOS I,et al.Semantic image segmentation with deep convolutional nets and fully connected CRFs[C]//Proceedings of International Conference on Learning Representation.San Diego,USA:[s.n.],2015:1-5.
[15] CHEN L C,PAPANDREOU G,KOKKINOS I,et al.DeepLab:semantic image segmentation with deep convolutional nets,atrous convolution,and fully connected CRFs[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2016,40(4):834-848.
[16] CHEN L C,PAPANDREOU G,SCHROFF F,et al.Rethinking atrous convolution for semantic image segmentation[EB/OL].(2017-12-05)[2020-04-22].https://arxiv.org/pdf/1706.05587.pdf.
[17] CHEN L C,ZHU Y,PAPANDREOU G,et al.Encoder-decoder with atrous separable convolution for semantic image segmentation[C]//Proceedings of European Conference on Computer Vision.Berlin,Germany:Springer,2018:833-851.
[18] HE K,GKIOXARI G,DOLLÁR P,et al.Mask R-CNN[C]//Proceedings of 2017 IEEE International Conference on Computer Vision.Washington D.C.,USA:IEEE Press,2017:2961-2969.
[19] LIN T Y,DOLLÁR P,GIRSHICK R,et al.Feature pyramid networks for object detection[EB/OL].(2017-04-19)[2020-04-22].https://arxiv.org/pdf/1612.03144.pdf.
[20] HE Kaiming,ZHANG Xiangyu,REN Shaoqing,et al.Deep residual learning for image recognition[C]//Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C.,USA:IEEE Press,2016:770-778.
[21] XIE S,GIRSHICK R,DOLLÁR P,et al.Aggregated residual transformations for deep neural networks[EB/OL].(2017-04-11)[2020-04-22].https://arxiv.org/pdf/1611. 05431.pdf.

选择文件类型/文献管理软件名称

选择包含的内容