面向嵌入式设备的高实时微小目标跟踪检测方法

doi:10.19678/j.issn.1000-3428.0060220

摘要/Abstract

摘要： 传统主流目标检测算法在嵌入式平台无法兼具高实时性与高准确性，难以应用于边缘智能等领域。为解决微小目标跟踪检测在嵌入式平台实时应用的瓶颈，提出一种高实时微小目标跟踪检测方法。利用轻量化神经网络的骨干网络和路径聚合网络，对整体网络进行针对化的剪枝优化，同时深度融合相关滤波算法，提升针对微小目标跟踪检测的准确度和速度。在3D物体场景渲染器自建的军事微小目标数据集上的实验结果表明，在100像素的极小目标跟踪识别中，与DarkNet53-CSP方法相比，该方法检测精度大幅提高，在400~10 000像素的微小目标识别跟踪中，检测精度与检测速度优于DarkNet53和ResNeXt50+CSP等算法。

关键词: 目标检测, 嵌入式平台, 多路影像, 卷积神经网络, 滤波算法

Abstract: Traditional mainstream target detection algorithms cannot provide both high real-time performance and high accuracy on embedded platforms, and thus have limited application in the edge intelligence field.In order to solve this real-time application bottleneck on embedded platforms, a highly real-time small target tracking and detection method is proposed.This method employs the backbone network of a lightweight neural network and a path aggregation network to prune the overall network for optimization.At the same time, it deeply integrates correlation filtering algorithms to improve the accuracy and speed of small target tracking and detection.The method is tested on a self-made small military target dataset built by using the 3D object and scene renderer.The experimental results show that in the case of 100 pixel small target tracking and recognition, the proposed method greatly improves the detection accuracy compared with Darknet53-CSP.In the case of 400~10 000 pixel small target tracking and recognition, the proposed method exhibits higher detection accuracy and speed than DarkNet53 and ResNeXt50+CSP.

Key words: target detection, embedded platforms, multi-channel image, convolutional neural network, filtering algorithm

中图分类号:

TP391

冒睿瑞, 江波. 面向嵌入式设备的高实时微小目标跟踪检测方法[J]. 计算机工程, 2021, 47(12): 291-298.

MAO Ruirui, JIANG Bo. Highly Real-Time Method of Small Target Tracking and Detection for Embedded Device[J]. Computer Engineering, 2021, 47(12): 291-298.

http://www.ecice06.com/CN/Y2021/V47/I12/291

图/表 12

20211214182930

20211214182937

20211214182940

20211214182944

20211214182948

20211214182952

20211214182956

20211214182959

20211214183003

20211214183006

20211214183009

20211214183013

参考文献

[1] WOMG A, SHAFIEE M J, LI F, et al.Tiny SSD:a tiny single-shot detection deep convolutional neural network for real-time embedded object detection[C]//Proceedings of the 15th Conference on Computer and Robot Vision.Washington D.C., USA:IEEE Press, 2018:95-101.
[2] 邵宇辉.基于背景建模的微小运动目标检测算法研究[D].西安:西安理工大学, 2019. SHAO Y H.Research on small moving target detection based on background modeling[D].Xi'an:Xi'an University of Technology, 2019.(in Chinese)
[3] CHEN C, LIU M Y, TUZEL O, et al.R-CNN for small object detection[C]//Proceedings of ACCVʼ16.Berlin, Germany:Springer, 2016:214-230.
[4] LI J, LIANG X, WEI Y, et al.Perceptual generative adversarial networks for small object detection[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2017:1222-1230.
[5] 吴言枫, 王延杰, 孙海江, 等.复杂动背景下的"低小慢"目标检测技术[J].中国光学, 2019, 12(4):854-866. WU Y F, WANG Y J, SUN H J, et al.LSS-target detection in complex sky backgrounds[J].Chinese Optics, 2019, 12(4):854-866.(in Chinese)
[6] 葛宝义, 左宪章, 胡永江.视觉目标跟踪方法研究综述[J].中国图象图形学报, 2018, 23(8):1091-1107. GE B Y, ZUO X Z, HU Y J.Review of visual object tracking technology[J].Journal of Image and Graphics, 2018, 23(8):1091-1107.(in Chinese)
[7] 胡昭华, 黄嘉净.基于快速多尺度估计的重新检测目标跟踪算法[J].微电子学与计算机, 2020, 37(9):11-17. HU Z H, HUANG J J.Re-detection target tracking algorithm based on fast multi-scale estimation[J].Microelectronics & Computer, 2020, 37(9):11-17.(in Chinese)
[8] BOLME D S, BERERIDGE J R, DRAPER B A, et al.Visual object tracking using adaptive correlation filters[C]//Proceedings of 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2010:2544-2550.
[9] 周勇.改进的卡尔曼滤波在目标跟踪中的应用[D].哈尔滨:哈尔滨工程大学, 2016. ZHOU Y.Application of the improved Kalman filter in target tracking[D].Harbin:Harbin Engineering University, 2016.(in Chinese)
[10] 韦越, 陈世超, 朱凤华, 等.基于稀疏卷积神经网络的模型剪枝方法[J].计算机工程, 2021, 47(10):61-66. WEI Y, CHEN S C, ZHU F H, et al.Model pruning method based on sparse convolutional neural network[J].Computer Engineering, 2021, 47(10):61-66.(in Chinese)
[11] TRONG T D, NGUYEN V T, TRAN M T.Lightweight deep convolutional network for tiny object recognition[C]//Proceedings of ICPRAMʼ18.Washington D.C., USA:IEEE Press, 2018:675-682.
[12] 杨德胜, 马冬.基于轻量级深层神经网络的人员安全管控平台研究[J].电力信息与通信技术, 2019, 17(6):1-7. YANG D S, MA D.Research on personnel safety management and control platform based on lightweight deep neural network[J].Electric Power Information and Communication Technology, 2019, 17(6):1-7.(in Chinese)
[13] 何洋.时空域快速自适应视频前景检测与去噪算法研究[D].西安:西安电子科技大学, 2017. HE Y.Research on fast adaptive video foreground detection and denoising algorithm in spatio-temporal domain[D].Xi'an:Xidian University, 2017.(in Chinese)
[14] 卢海伟, 袁晓彤.基于层融合特征系数的动态网络结构化剪枝[J].模式识别与人工智能, 2019, 32(11):1051-1059. LU H W, YUAN X D.Dynamic network structured pruning via feature coefficients of layer fusion[J].Pattern Recognition and Artificial Intelligence, 2019, 32(11):1051-1059.(in Chinese)
[15] 尹宏鹏, 陈波, 柴毅, 等.基于视觉的目标检测与跟踪综述[J].自动化学报, 2016, 42(10):1466-1489. YIN H P, CHEN B, CAI Y, et al.Vision-based object detection and tracking:a review[J].Acta Automatica Sinica, 2016, 42(10):1466-1489.(in Chinese)
[16] 吴天舒, 张志佳, 刘云鹏, 等.基于改进SSD的轻量化小目标检测算法[J].红外与激光工程, 2018, 47(7):47-53. WU T S, ZHANG Z J, LIU Y P, et al.A lightweight small object detection algorithm based on improved SSD[J].Infrared and Laster Engineering, 2018, 47(7):47-53.(in Chinese)
[17] LIU S, QI L, QIN H, et al.Path aggregation network for instance segmentation[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2018:8759-8768.
[18] 赵杰.基于相关滤波的多策略融合运动目标跟踪算法研究[D].昆明:昆明理工大学, 2019. ZHAO J.Research on multi-strategy fusion moving target tracking algorithm based on correlation filtering[D].Kunming:Kunming University of Science and Technology, 2019.(in Chinese)
[19] 王明丽, 王刚, 郭晓新, 等.基于跨层复制连接卷积神经网络的遥感图像融合[J].吉林大学学报(理学版), 2020, 58(4):913-922. WANG M L, WANG G, GUO X X, et al.Remote sensing image fusion based on cross layer copy connection convolutional neural network[J].Journal of Jilin University Science Edition, 2020, 58(4):913-922. (in Chinese)
[20] 林成创, 赵淦森, 尹爱华, 等.AS-PANet:改进路径增强网络的重叠染色体实例分割[J].中国图象图形学报, 2020, 25(10):2271-2280. LIN C C, ZHAO J S, YIN A H, et al.AS-PANet:improved overlapping chromosome instance segmentation of path enhancement network[J].Journal of Image and Graphics, 2020, 25(10):2271-2280.(in Chinese)
[21] 杨元挺.基于粒子滤波的灰度目标跟踪算法[J].厦门大学学报(自然科学版), 2012, 51(1):33-36. YANG Y T.A gray object tracking algorithm based on particle filter[J].Journal of Xiamen University(Natural Science), 2012, 51(1):33-36.(in Chinese)
[22] IOFFE S, SZEGEDY C.Batch normalization:accelerating deep network training by reducing internal covariate shift[C]//Proceedings of the 32nd International Conference on Machine Learning.Washington D.C., USA:IEEE Press, 2015:448-456.

选择文件类型/文献管理软件名称

选择包含的内容