[1]蔡英凤,王海,孙晓强,等.基于深度模型的场景自适应行人检测[J].东南大学学报(自然科学版),2017,47(4):679-684.[doi:10.3969/j.issn.1001-0505.2017.04.009]
 Cai Yingfeng,Wang Hai,Sun Xiaoqiang,et al.Scene adaptive pedestrian detection algorithm based on deep model[J].Journal of Southeast University (Natural Science Edition),2017,47(4):679-684.[doi:10.3969/j.issn.1001-0505.2017.04.009]
点击复制

基于深度模型的场景自适应行人检测()
分享到:

《东南大学学报(自然科学版)》[ISSN:1001-0505/CN:32-1178/N]

卷:
47
期数:
2017年第4期
页码:
679-684
栏目:
计算机科学与工程
出版日期:
2017-07-20

文章信息/Info

Title:
Scene adaptive pedestrian detection algorithm based on deep model
作者:
蔡英凤1王海2孙晓强1袁朝春1陈龙1江浩斌2
1江苏大学汽车工程研究院, 镇江 212013; 2江苏大学汽车与交通工程学院, 镇江 212013
Author(s):
Cai Yingfeng1 Wang Hai2 Sun Xiaoqiang1 Yuan Chaochun1 Chen Long1 Jiang Haobin2
1Automotive Engineering Research Institute, Jiangsu University, Zhenjiang 212013, China
2School of Automotive and Traffic Engineering, Jiangsu University, Zhenjiang 212013, China
关键词:
场景自适应 行人检测 深度结构 卷积神经网络
Keywords:
scene adaption pedestrian detection deep structure deep convolutional neural network
分类号:
TP391.4
DOI:
10.3969/j.issn.1001-0505.2017.04.009
摘要:
针对现有基于机器学习的行人检测算法存在当训练样本和目标场景样本分布不匹配时检测效果显著下降的缺陷,提出一种基于深度模型的场景自适应行人检测算法.首先,受Bagging机制启发,以相对独立源数据集构建多个分类器,再通过投票实现带置信度度量的样本自动选取;其次,利用DCNN深度结构的特征自动抽取能力,加入一个自编码器对源-目标场景下特征相似度进行度量,提出了一种基于深度模型的场景自适应分类器模型并设计了训练方法.在KITTI数据库的测试结果表明,所提算法较现有非场景自适应行人检测算法具有较大的优越性;与已有的场景自适应学习算法相比较,该算法在检测率上平均提升约4%.
Abstract:
To solve the problem that the detection effect of the existing machine learning based pedestrian detection algorithms decreases dramatically when the distributions of training samples and scene target samples do not match, a scene adaptive pedestrian detection algorithm based on the deep model is proposed. First, inspired by the Bagging(Bootstrap aggregating)mechanism, multiple relatively independent source samples are used to build multiple classifiers and then target training samples with confidence score are generated by voting. Secondly, using the automatic feature extraction ability of DCNN(deep convolutional neural network)and adding a deep auto-encoder to perform the source-target scene feature similarity calculation, a deep model-based scene adaptive classifier model is proposed and its training algorithm is designed. The experiments on the KITTI dataset demonstrate that the proposed algorithm performs better than the existing non-scene adaptive pedestrian detection algorithms. Besides, compared with the existing scene adaptive object detection algorithms, the proposed algorithm improves the detection rate on average by approximately 4%.

参考文献/References:

[1] 曾波波, 王贵锦, 林行刚. 基于颜色自相似度特征的实时行人检测[J]. 清华大学学报(自然科学版), 2012, 52(4): 571-574.
  Zeng Bobo,Wang Guijin,Lin Xinggang. Color self-similarity feature based real-time pedestrian detection[J]. Journal of Tsinghua University(Science and Technology),2012, 52(4): 571-574.(in Chinese)
[2] 姚雪琴, 李晓华, 周激流. 基于边缘对称性和HOG的行人检测方法[J]. 计算机工程, 2012, 38(5): 179-182. DOI:10.3969/j.issn.1000-3428.2012.05.055.
Yao Xueqin, Li Xiaohua, Zhou Jiliu. Pedestrian detection method based on edge symmetry and HOG[J]. Computer Engineering, 2012, 38(5): 179-182. DOI:10.3969/j.issn.1000-3428.2012.05.055. (in Chinese)
[3] Hoang V D, Le M H, Jo K H. Hybrid cascade boosting machine using variant scale blocks based HOG features for pedestrian detection[J]. Neurocomputing, 2014, 135: 357-366. DOI:10.1016/j.neucom.2013.12.017.
[4] 孙锐, 陈军, 高隽. 基于显著性检测与HOG-NMF特征的快速行人检测方法[J]. 电子与信息学报, 2013,35(8): 1921-1926. DOI:10.3724/SP.J.1146.2012.01700.
Sun Rui, Chen Jun, Gao Jun. Fast pedestrian detection based on saliency detection and HOG-NMF features[J]. Journal of Electronics & Information Technology, 2013, 35(8): 1921-1926. DOI:10.3724/SP.J.1146.2012.01700. (in Chinese)
[5] 孙锐, 侯能干, 陈军. 基于特征融合和交叉核SVM的快速行人检测方法[J]. 光电工程, 2014, 41(2): 53-62. DOI:10.3969/j.issn.1003-501X.2014.02.009.
Sun Rui, Hou Nenggan, Chen Jun. Fast pedestrian detection method based on features fusion and intersection kernel SVM[J]. Opto-Electronic Engineering, 2014, 41(2): 53-62. DOI:10.3969/j.issn.1003-501X.2014.02.009. (in Chinese)
[6] 崔华, 张骁, 郭璐, 等. 多特征多阈值级联 AdaBoost 行人检测器[J]. 交通运输工程学报, 2015, 15(2): 109-117. DOI:10.3969/j.issn.1671-1637.2015.02.014.
Cui Hua, Zhang Xiao, Guo Lu, et al. Cascade AdaBoost pedestrian detector with multi-features and multi-thresholds[J]. Journal of Traffic and Transportation Engineering, 2015, 15(2): 109-117. DOI:10.3969/j.issn.1671-1637.2015.02.014. (in Chinese)
[7] 王海, 蔡英凤, 陈龙, 等. 基于 Haar-NMF 特征和改进 SOMPNN 的车辆检测算法[J]. 东南大学学报(自然科学版), 2016, 46(3): 499-504. DOI:10.3969/j.issn.1001-0505.2016.03.008.
Wang Hai, Cai Yingfeng, Chen Long, et al. Vehicle detection algorithm based on Haar-NMF features and improved SOMPNN[J]. Journal of Southeast University(Natural Science Edition), 2016, 46(3): 499-504. DOI:10.3969/j.issn.1001-0505.2016.03.008. (in Chinese)
[8] 蔡英凤, 王海. 视觉车辆识别迁移学习算法[J]. 东南大学学报(自然科学版), 2015, 45(2): 275-280.DOI:10.3969/j.issn.1001-0505.2015.02.015.
Cai Yingfeng, Wang Hai. Vision based vehicle detection transfer learning algorithm[J]. Journal of Southeast University(Natural Science Edition), 2015, 45(2): 275-280. DOI:10.3969/j.issn.1001-0505.2015.02.015. (in Chinese)
[9] Daumé H Ⅲ, Marcu D. Domain adaptation for statistical classifiers[J]. Journal of Artificial Intelligence Research,2011, 26(1):101-126.
[10] Pan S J, Yang Q. A survey on transfer learning[J]. IEEE Transactions on Knowledge and Data Engineering, 2010, 22(10): 1345-1359. DOI:10.1109/tkde.2009.191.
[11] 陈荣, 曹永锋, 孙洪. 基于主动学习和半监督学习的多类图像分类[J]. 自动化学报, 2011, 37(8): 954-962.
  Chen Rong, Cao Yongfeng, Sun Hong. Multi-class image classification with active learning and semi-supervised learning[J]. Acta Automatica Sinica, 2011, 37(8): 954-962.(in Chinese)
[12] Wang M, Li W, Wang X. Transferring a generic pedestrian detector towards specific scenes[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Providence, Rhode Island, USA, 2012:3274-3281.
[13] Liang F, Tang S, Wang Y, et al. A sparse coding based transfer learning framework for pedestrian detection[C]//Proceedings of International Conference on Multimedia Modeling. Huangshan, China, 2013: 272-282.
[14] Nair V, Clark J J. An unsupervised online learning framework for moving object detection[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Washington, DC, USA, 2004: 317-324. DOI:10.1109/cvpr.2004.1315181.
[15] Rosenberg C, Hebert M, Schneiderman H. Semi-supervised self-training of object detection models[C]//Proceedings of IEEE Workshop on Application of Computer Vision. Breckenridge, CO, USA, 2005: 29-36. DOI:10.1109/acvmot.2005.107.
[16] Wang M, Wang X. Automatic adaptation of a generic pedestrian detector to a specific traffic scene[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Colorado, CO, USA, 2011:3401-3408.
[17] Sharma P,Huang C,Nevatia R. Unsupervised incremental learning for improved object detection in a video[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Providence, Rhode Island, USA, 2012: 3298-3305.
[18] Li X, Ye M, Fu M, et al. Domain adaption of vehicle detector based on convolutional neural networks[J]. International Journal of Control, Automation and Systems, 2015, 13(4): 1020-1031. DOI:10.1007/s12555-014-0119-z.
[19] Cao X, Wang Z, Yan P, et al. Transfer learning for pedestrian detection[J]. Neurocomputing, 2013, 100: 51-57. DOI:10.1016/j.neucom.2011.12.043.
[20] Hamid P, Sajad P, Zahra R, et al. CDEBMTE: Diverse ensemble based on manipulation of training example[J]. Pattern Recognition, 2012, 7329: 197-206.
[21] 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. DOI:10.1109/TPAMI.2009.167.
[22] Krizhevsky A, Sutskever I, Hinton G. Imagenet classification with deep convolutional neural networks[C]//Advances in Neural Information Processing Systems. Lake Tahoe, USA, 2012: 1097-1105.
[23] Geiger A, Lenz P, Urtasun R. Are we ready for autonomous driving? The KITTI vision benchmark suite[C]//2012 IEEE Conference on Computer Vision and Pattern Recognition. Providence, Rhode Island, USA, 2012: 3354-3361. DOI:10.1109/cvpr.2012.6248074.
[24] Tomè D, Monti F, Baroffio L, et al. Deep convolutional neural networks for pedestrian detection[J]. Signal Processing: Image Communication, 2016, 47: 482-489. DOI:10.1016/j.image.2016.05.007.

备注/Memo

备注/Memo:
收稿日期: 2016-11-06.
作者简介: 蔡英凤(1985—),女,博士,副教授,caicaixiao0304@126.com.
基金项目: 国家自然科学基金资助项目(U1564201,61403172,61601203)、中国博士后基金资助项目(2014M561592,2015T80511)、江苏省重点研发计划资助项目(BE2016149)、江苏省自然科学基金资助项目(BK20140555)、江苏省六大人才高峰资助项目(2014-DZXX-040,2015-JXQC-012).
引用本文: 蔡英凤,王海,孙晓强,等.基于深度模型的场景自适应行人检测[J].东南大学学报(自然科学版),2017,47(4):679-684. DOI:10.3969/j.issn.1001-0505.2017.04.009.
更新日期/Last Update: 2017-07-20