基于少量样本学习的多目标检测跟踪方法

doi:10.12263/DZXB.20180045

电子学报 ›› 2021, Vol. 49 ›› Issue (1): 183-191.DOI: 10.12263/DZXB.20180045

基于少量样本学习的多目标检测跟踪方法

罗大鹏¹, 杜国庆¹, 曾志鹏¹, 魏龙生², 高常鑫³, 陈应⁴, 肖菲⁴, 罗琛⁵

1. 中国地质大学(武汉)机械与电子信息学院, 湖北武汉 430074;
2. 中国地质大学(武汉) 自动化学院, 湖北武汉 430074;
3. 华中科技大学自动化学院, 湖北武汉 430074;
4. 中建三局智能技术有限公司, 湖北武汉 430070;
5. 北京师范大学惠州附属学校, 广东惠州 516002

收稿日期:2018-11-26 修回日期:2019-01-09 出版日期:2021-01-25
- 通讯作者:
- 罗大鹏
- 作者简介:
- 曾志鹏 男,1993年生,湖北仙桃人,中国地质大学(武汉)机械与电子信息学院硕士研究生.主要研究方向为计算机视觉、机器学习.E-mail:zengzhipeng@cug.edu.cn;杜国庆 男,1993年生,湖北武汉人,中国地质大学(武汉)机械与电子信息学院硕士研究生.主要研究方向为发育神经网络、自主学习、目标跟踪与检测.E-mail:guoqingdu@foxmail.com
- 基金资助:
- 国家自然科学基金 (No.61302137，No.61603357，No.61603354）; 中央高校基本科研业务费专项 (No.CUGL170210）

Multi-Object Detection and Tracking Based on Few-Shot Learning

LUO Da-peng¹, DU Guo-qing¹, ZENG Zhi-peng¹, WEI Long-sheng², GAO Chang-xin³, CHENG Ying⁴, XIAO Fei⁴, LUO Chen⁵

1. School of Electronic Information and Mechanics, China University of Geosciences, Wuhan, Hubei 430074, China;
2. School of Automation, China University of Geosciences, Wuhan, Hubei 430074, China;
3. School of Automation, Huazhong University of Science and Technology, Wuhan, Hubei 430074, China;
4. Intelligent Technology Co., Ltd. of Chinese Construction Third Engineering Bureau, Wuhan, Hubei 430070, China;
5. Huizhou School Affiliated to Beijing Normal University, Huizhou, Guangdong 516002, China

Received:2018-11-26 Revised:2019-01-09 Online:2021-01-25 Published:2021-01-25
- Corresponding author:
- LUO Da-peng
- Supported by:
- National Natural Science Foundation of China (No.61302137, No.61603357, No.61603354); Fundamental Research Funds for the Central Universities (No.CUGL170210)

摘要/Abstract

摘要： 视频目标检测跟踪算法一直是计算机视觉领域的研究热点，目前大部分方法均需人工采集样本训练检测模型，搭建目标检测跟踪系统.当目标成像条件发生变化时，需重新采集样本，训练模型，调试整个检测跟踪系统，耗费大量人力、物力.本文提出一种基于少量样本学习的多目标检测跟踪算法，只需在监控视频第一帧指定待检测目标，即可自主生成混合分类模型，进行目标检测.采用在线渐进学习算法学习目标姿态变化，更新该模型.结合基于颜色的目标跟踪算法，自动构建高精度目标检测跟踪系统.整个过程无需手工采集、标注训练样本.因此，易于扩展到其它监控场景，通过自主学习形成该场景专用的检测跟踪系统，实现不同监控环境下，不同成像条件下都有较好的检测跟踪效果.实验表明，本方法能自主学习多种监控场景下的目标姿态，无需手工标注训练样本，在基于在线学习的算法中有最佳的检测精度，此外也取得了和离线目标检测跟踪系统相似的性能.

关键词: 少量样本学习, 多目标检测, 多目标跟踪, 在线学习

Abstract: Video object detection and tracking algorithms have become the research focus in the field of computer vision. Traditional methods need to manually collect samples to train detection models, and build object detection and tracking systems. When the imaging conditions change, it is necessary to re-collect samples to train the detection model and re-adjust the entire detection and tracking system, which requires tedious human efforts. In this paper, a multi-object detection and tracking algorithm is proposed based on few-shot learning. With this approach, a hybrid classifier that models one object class can be generated by simply marking several bounding boxes around the object in the first video frame. An online gradual learning algorithm is proposed to learn the object pose changes and update the model. Combined with the color-based object tracking algorithm, our method automatically builds high-precision object detection and tracking systems without manual collection and labeling training samples. This approach can be conveniently replicated many times in different surveillance scenes and produce scene-specific detectors under various camera viewpoints. Experimental results on several video datasets show our approach achieves comparable performance to robust supervised methods, and outperforms the state-of-the-art online learning methods in varying imaging conditions.

Key words: few-shot learning, multi-object detection, multi-object tracking, online learning

中图分类号:

TP391

罗大鹏, 杜国庆, 曾志鹏, 等. 基于少量样本学习的多目标检测跟踪方法[J]. 电子学报, 2021, 49(1): 183-191.

LUO Da-peng, DU Guo-qing, ZENG Zhi-peng, et al. Multi-Object Detection and Tracking Based on Few-Shot Learning[J]. Acta Electronica Sinica, 2021, 49(1): 183-191.

参考文献

[1] Gwak J.Multi-object tracking through learning relational appearance features and motion patterns[J].Computer Vision and Image Understanding,2017,162(9):103-115
[2] Felzenszwalb P F,Girshick R B,Mcallester D,et al.Object detection with discriminatively trained part based models[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2010,32(9):1627-1645.
[3] Bourdev L,J Malik,Poselets.Body part detectors trained using 3d human pose annotations[A].2009 IEEE 12th International Conference on Computer Vision[C].Kyoto,Japan:IEEE,2010.1365-1372.
[4] Yang Y,Shu G,Shah M.Semi-supervised Learning of Feature Hierarchies for Object Detection in a Video[A].2013 IEEE Conference onComputer Vision and Pattern Recognition[C].Portland,USA:IEEE,2013.1650-1657
[5] Wang X,Li W,Wang M.Transferring a generic pedestrian detector towards specific scenes[A].2012 IEEE Conference on Computer Vision and Pattern Recognition[C].Providence,RI,USA:IEEE,2012.3274-3281.
[6] Cinbis R G,Verbeek J,et al.Weakly supervised object localization with multi-fold multiple instance learning[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2016,39(1):189-203.
[7] Liu Y,Wang Y,et al.Soft hough foresterts:Generalized hough transform based object detection from soft-labelled training data[J].Pattern Recognition,2016,60:145-156.
[8] Singh K K,Xiao F,Yong J L.Track and transfer:Watching videos to simulate strong human supervision for weakly-supervised object detection[A].Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition[C].LasVegas,USA:IEEE,2016.3548-3556.
[9] Qi Z,Xu Y,Wang L,Song Y.Online multiple instance boosting for object detection[J].Neur-ocomputing,2011,74(10):1769-1775.
[10] Nair Clark V,J J.An unsupervised online learning frame-work for moving object detection[A].IEEE Computer Society Conference on Computer Vision and Pattern Recognition[C].Washington,DC,USA:IEEE,2004.317-325.
[11] Wang X,Hua G,Han T X.Detection by detections:Non-parametric detector adaptation for a video[A]. Computer Vision and Pattern Recognition[C].Providence,RI,USA:IEEE,2012.350-357.
[12] Sharma P,Nevatia R.Ecient detector adaptation for ob-ject detection in a video[A].Computer Vision and Pattern Recognition[C].Kyoto,Japan:IEEE,2013.3254-3261.
[13] Wozniak M,Grana M,et al.A survey of multiple classifier systems as hybrid systems[J].Information Fusion,2014,16(1):3-17.
[14] Villamizar M,Andrade-Cetto J,Sanfeliu A,et al.Boosted random ferns for object detection[J].IEEE Transactions on Pattern Analysis Machine Intelligence,2018,40(2):272-288.
[15] Ozuysal M,Calonder M,et al.Fast keypoint recognition using random ferns[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2010,32(3):448-461.
[16] 罗大鹏,罗琛,魏龙生,韩家宝,王勇,马丽.基于在线随机蕨分类器的实时视觉感知系统[J].电子学报,2016,44(5):1139-1148. LUO Da-peng,LUO Chen,WEI Long-sheng,HAN Jia-bao,WANG Yong,MA Li.Real time visual perception system based on online fern classifier[J].Acta Electronica Sinica,2016,44(5):1139-1148.(in Chinese)
[17] Kuhn H.W.Variants of the hungarian method for assignment problems[J].Naval Research Logistics,2010,3(4):253-258.
[18] CAVIARDATA[D/OL].http://groups.inf.ed.ac.uk/vision/CAVIAR/,2018.
[19] Dollár P,Appel R,et al.Fast feature pyramids for object detection[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2014,36(8):1532-1545.
[20] Ferryman J,Shahrokni A.PETS2009:Dataset and challenge[A].Proceedings of 2009 Twelfth IEEE International Workshop on Performance Evaluation of Tracking and Surveillance[C].Snowbird,USA:IEEE,2009.1-6.
[21] Andriyenko A,Schindler K.Multi-target tracking by continuous energy minimization[A].Computer Vision and Pattern Recognition[C].Colorado Springs,USA:IEEE,2011.1265-1272.
[22] Fragkiadaki K,Zhang W,et al.Two-granularity tracking:mediating trajectory and detection graphs for tracking under occlusions[A].European Conference on Computer Vision[C].Florence,Italy:Springer,2012.552-565.
[23] Milan A,Schindler K,Roth S.Detection and trajectory level exclusion in multiple object tracking[A].IEEE Conference on Computer Vision and Pattern Recognition[C].Portland,USA:IEEE,2013. 3682-3689.
[24] Milan A,Leal-Taixé L,Schindler K,Reid I.Joint tracking and segmentation of multiple targets[A].IEEE Conference on Computer Vision and Pattern Recognition[C].Boston,USA:IEEE,2015.5397-5406.
[25] Roth P M,Grabner H,et al.Online conservative learning for person detection[A].Iternational Conference on Computer Communications and Networks[C].Beijing,China:IEEE,2005.223-230.

基于少量样本学习的多目标检测跟踪方法

Multi-Object Detection and Tracking Based on Few-Shot Learning

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

[1]	陆克中, 陈超凡, 蔡桓, 吴定明. 面向概念漂移和类不平衡数据流的在线分类算法[J]. 电子学报, 2022, 50(3): 585-597.
[2]	侯利明, 连峰, 谭顺成, 徐从安. 闪烁噪声统计特性未知情况下的鲁棒广义标签多伯努利滤波器[J]. 电子学报, 2021, 49(7): 1346-1353.
[3]	陈辉, 贺忠良, 邓东明, 李国财. 高斯混合多伯努利滤波器基于柯西施瓦兹散度的传感器控制方法[J]. 电子学报, 2020, 48(4): 706-716.
[4]	陈辉, 贺忠良, 连峰, 黎慧波. 基于高斯混合多目标滤波器的传感器控制策略[J]. 电子学报, 2019, 47(3): 521-530.
[5]	连峰, 张修立, 魏博, 侯利明, 韩崇昭, 王伟. 基于多目标MS-OSPA下界的分散式传感器选择[J]. 电子学报, 2019, 47(10): 2158-2165.
[6]	魏波, 吴瑞峰, 张文生, 吕敬钦, 王莹莹, 夏学文. 具有双重稀疏机制的在线学习算法[J]. 电子学报, 2019, 47(10): 2202-2210.
[7]	熊志刚, 黄树彩, 苑智玮, 赵炜. 三维空间纯方位多目标跟踪PHD算法[J]. 电子学报, 2018, 46(6): 1371-1377.
[8]	曲长文, 冯奇, 李廷军. 基于分治-贪心算法的高斯混合多观测站CPHD滤波器[J]. 电子学报, 2018, 46(10): 2472-2479.
[9]	罗大鹏, 罗琛, 魏龙生, 韩家宝, 王勇, 马丽. 基于在线随机蕨分类器的实时视觉感知系统[J]. 电子学报, 2016, 44(5): 1139-1148.
[10]	李康, 何发智, 潘一腾, 孙航. 基于簇相似的多分类器目标跟踪算法[J]. 电子学报, 2016, 44(4): 821-825.
[11]	徐从安, 熊伟, 刘瑜, 何友. 新生目标强度未知的单量测PHD滤波器[J]. 电子学报, 2016, 44(10): 2300-2307.
[12]	刘宗香, 谢维信, 李丽娟, 李良群. 非线性高斯系统边缘分布多目标贝叶斯滤波器[J]. 电子学报, 2015, 43(9): 1689-1695.
[13]	姜明新, 王培昌, 王洪玉. 基于视频多目标跟踪的高度测量算法[J]. 电子学报, 2015, 43(3): 591-596.
[14]	权伟, 陈锦雄, 余南阳. 在线学习多重检测的可视对象跟踪方法[J]. 电子学报, 2014, 42(5): 875-882.
[15]	刘宗香, 谢维信, 王品, 余友. 一种具有信息保持能力的GM-PHD滤波器[J]. 电子学报, 2013, 41(8): 1603-1608.