·1104· 智能系统学报 第16卷 图。同时，在自顶向下和自底向上的双向融合中 [10]SINGH B.DAVIS L S.An analysis of scale invariance 加入了通道注意力机制，提高了特征融合的效 in object detection-SNIP[C]//Proceedings of 2018 率。最后，针对目标的漏检情况，本文提出了一 IEEE/CVF Conference on Computer Vision and Pattern 种改进的正负样本判定策略，提取到被检测目标 Recognition.Salt Lake City,USA,2018:3578-3587. 更多的局部特征。经过实验对比，本文所提出的 [11]温静，李雨萌.基于多尺度反卷积深度学习的显著性 算法模型相较于传统S$D算法，平均准确率方面 检测)计算机科学，2020,4711)：179-185 提高3.1%.表明了本文所提算法的有效性。 WEN Jing,LI Yumeng.Salient object detection based 参考文献： on multi-scale deconvolution deep learning[J].Com- puter science,2020,47(11):179-185. [1]GIRSHICK R.DONAHUE J.DARRELL T.et al.Rich [12]LI Jianan,LIANG Xiaodan,WEI Yunchao,et al.Per- feature hierarchies for accurate object detection and se- ceptual generative adversarial networks for small object mantic segmentation[C]//Proceedings of the IEEE Con- detection[C]//Proceedings of the IEEE Conference on ference on Computer Vision and Pattern Recognition. Computer Vision and Pattern Recognition.Honolulu, Columbus,USA,2014:580-587. USA.2017:1951-1959 [2]UIJLINGS J RR.VAN DE SANDE K EA.GEVERS T. [13]GOODFELLOW I J,POUGET-ABADIE J,MIRZA M. et al.Selective search for object recognition[J].Interna- et al.Generative adversarial nets[Cl/Proceedings of the tional journal of computer vision,2013,104(2):154-171. 27th International Conference on Neural Information [3]HE Kaiming,ZHANG Xiangyu,REN Shaoqing,et al. Processing Systems.Montreal,Canada,2014: Spatial pyramid pooling in deep convolutional networks 2672-2680 for visual recognition[J].IEEE transactions on pattern [14]LIN T Y,DOLLAR P,GIRSHICK R,et al.Feature pyr- analysis and machine intelligence,2015,37(9): amid networks for object detection[C]//Proceedings of 19041916. the IEEE Conference on Computer Vision and Pattern [4]GIRSHICK R.Fast R-CNN[C]//Proceedings of the IEEE Recognition.Honolulu,USA,2017:936-944 International Conference on Computer Vision.Santiago, [15]刘涛，汪西莉.采用卷积核金字塔和空洞卷积的单阶 Chile,.2015:1440-1448. 段目标检测[].中国图象图形学报，2020,25(1) [5]REN Shaoqing,HE Kaiming,GIRSHICK R,et al.Faster 102-112. R-CNN:towards real-time object detection with region LIU Tao,WANG Xili.Single-stage object detection us- proposal networks[J].IEEE transactions on pattern ana- ing filter pyramid and atrous convolution[J].Journal of lysis and machine intelligence,2017,39(6):1137-1149. image and graphics,2020,25(1):102-112. [6]REDMON J.DIVVALA S,GIRSHICK R,et al.You [16]陈景明，金杰，王伟锋.基于特征金字塔网络的改进算 only look once:unified,real-time object detection[C]// 法[).激光与光电子学进展，2019,56(21)：211505. Proceedings of 2016 IEEE Conference on Computer Vis- CHEN Jingming,JIN Jie,WANG Weifeng.Improved ion and Pattern Recognition.Las Vegas,USA,2016: algorithm based on feature pyramid networks[J].Laser 779-788. &optoelectronics progress,2019,56(21):211505 [7]REDMON J.FARHADI A.YOLO9000:better,faster, [17刀张涛，张乐.一种基于多尺度特征融合的目标检测算 stronger[C]//Proceedings of 2017 IEEE Conference on 法[J.激光与光电子学进展，2021.58(2)：0215003 Computer Vision and Pattern Recognition.Honolulu, ZHANG Tao,ZHANG Le.Multiscale feature fusion- USA.2017:6517-6525 based object detection algorithm[J].Laser optoelec- [8]LIU Wei,ANGUELOV D,ERHAN D,et al.SSD:single tronics progress,2021,58(2):0215003. shot multibox detector[C]//Proceedings of the 14th [18]和超，张印辉，何自芬.多尺度特征融合工件目标语义 European Conference on Computer Vision.Amsterdam, 分割[J.中国图象图形学报，2020,25(3)：476-485. The Netherland,2016:21-37. HE Chao,ZHANG Yinhui,HE Zifen.Semantic seg- [9]FU Chengyang,LIU Wei,RANGA A,et al.DSSD:de- mentation of workpiece target based on multiscale fea- convolutional single shot detector[Cl//Proceedings of the ture fusion[J].Journal of image and graphics,2020, IEEE Conference on Computer Vision and Pattern Recog- 25(3:476-485. nition.Hawaii,USA.2017:2881-2890. [19]鞠默然，罗江宁，王仲博，等.融合注意力机制的多尺图。同时，在自顶向下和自底向上的双向融合中 加入了通道注意力机制，提高了特征融合的效 率。最后，针对目标的漏检情况，本文提出了一 种改进的正负样本判定策略，提取到被检测目标 更多的局部特征。经过实验对比，本文所提出的 算法模型相较于传统 SSD 算法，平均准确率方面 提高 3.1%，表明了本文所提算法的有效性。 参考文献： GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and se￾mantic segmentation[C]//Proceedings of the IEEE Con￾ference on Computer Vision and Pattern Recognition. Columbus, USA, 2014: 580−587. [1] UIJLINGS J R R, VAN DE SANDE K E A, GEVERS T, et al. Selective search for object recognition[J]. Interna￾tional journal of computer vision, 2013, 104(2): 154–171. [2] HE Kaiming, ZHANG Xiangyu, REN Shaoqing, et al. Spatial pyramid pooling in deep convolutional networks for visual recognition[J]. IEEE transactions on pattern analysis and machine intelligence, 2015, 37(9): 1904–1916. [3] GIRSHICK R. Fast R-CNN[C]//Proceedings of the IEEE International Conference on Computer Vision. Santiago, Chile, 2015: 1440−1448. [4] REN Shaoqing, HE Kaiming, GIRSHICK R, et al. Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE transactions on pattern ana￾lysis and machine intelligence, 2017, 39(6): 1137–1149. [5] REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: unified, real-time object detection[C]// Proceedings of 2016 IEEE Conference on Computer Vis￾ion and Pattern Recognition. Las Vegas, USA, 2016: 779−788. [6] REDMON J, FARHADI A. YOLO9000: better, faster, stronger[C]//Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu, USA, 2017: 6517−6525. [7] LIU Wei, ANGUELOV D, ERHAN D, et al. SSD: single shot multibox detector[C]//Proceedings of the 14th European Conference on Computer Vision. Amsterdam, The Netherland, 2016: 21−37. [8] FU Chengyang, LIU Wei, RANGA A, et al. DSSD: de￾convolutional single shot detector[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recog￾nition. Hawaii, USA, 2017: 2881−2890. [9] SINGH B, DAVIS L S. An analysis of scale invariance in object detection-SNIP[C]//Proceedings of 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City, USA, 2018: 3578−3587. [10] 温静, 李雨萌. 基于多尺度反卷积深度学习的显著性 检测 [J]. 计算机科学, 2020, 47(11): 179–185. WEN Jing, LI Yumeng. Salient object detection based on multi-scale deconvolution deep learning[J]. Com￾puter science, 2020, 47(11): 179–185. [11] LI Jianan, LIANG Xiaodan, WEI Yunchao, et al. Per￾ceptual generative adversarial networks for small object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Honolulu, USA, 2017: 1951−1959. [12] GOODFELLOW I J, POUGET-ABADIE J, MIRZA M, et al. Generative adversarial nets[C]//Proceedings of the 27th International Conference on Neural Information Processing Systems. Montreal, Canada, 2014: 2672−2680. [13] LIN T Y, DOLLÁR P, GIRSHICK R, et al. Feature pyr￾amid networks for object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Honolulu, USA, 2017: 936−944. [14] 刘涛, 汪西莉. 采用卷积核金字塔和空洞卷积的单阶 段目标检测 [J]. 中国图象图形学报, 2020, 25(1): 102–112. LIU Tao, WANG Xili. Single-stage object detection us￾ing filter pyramid and atrous convolution[J]. Journal of image and graphics, 2020, 25(1): 102–112. [15] 陈景明, 金杰, 王伟锋. 基于特征金字塔网络的改进算 法 [J]. 激光与光电子学进展, 2019, 56(21): 211505. CHEN Jingming, JIN Jie, WANG Weifeng. Improved algorithm based on feature pyramid networks[J]. Laser & optoelectronics progress, 2019, 56(21): 211505. [16] 张涛, 张乐. 一种基于多尺度特征融合的目标检测算 法 [J]. 激光与光电子学进展, 2021, 58(2): 0215003. ZHANG Tao, ZHANG Le. Multiscale feature fusion￾based object detection algorithm[J]. Laser & optoelec￾tronics progress, 2021, 58(2): 0215003. [17] 和超, 张印辉, 何自芬. 多尺度特征融合工件目标语义 分割 [J]. 中国图象图形学报, 2020, 25(3): 476–485. HE Chao, ZHANG Yinhui, HE Zifen. Semantic seg￾mentation of workpiece target based on multiscale fea￾ture fusion[J]. Journal of image and graphics, 2020, 25(3): 476–485. [18] [19] 鞠默然, 罗江宁, 王仲博, 等. 融合注意力机制的多尺 ·1104· 智 能 系 统 学 报 第 16 卷