基于局部深度匹配的行人再识别
|
摘要
针对行人再识别精度低的难题进行研究,提出了一种新的基于分块匹配的行人再识别方法。首先,引入带人体结构信息的人体DPM对行人外观进行分割,得到的带语义信息的身体部件作为匹配识别的基本单元;其次,基于深度神经网络模型提取各部件的深度特征作为匹配依据;再次,基于余弦距离判断各身体部件与目标行人对应部件的相似性;最后,融合所有身体部件的识别结果得到最终的再识别结果。实验结果表明,跟已有方法相比,该方法具有更好的鲁棒性,在识别精度上有较明显的优势。
This paper presented a new method based on part matching to improve the accuracy of person re-identification. First of all,a person DPM which carried the information of person structure was used to segment the human body into parts with semantic meanings,and these parts were used as basic units for re-identification. Secondly,the deep features of these body parts were extracted by deep neural network model. Thirdly,each body part of the testing person was compared with the corresponding body part of the target person based on the deep feature and cosine distance. Finally,the matching results from all the body parts were fused to make the final decision. Experimental results show that this method is more robust and it outperforms most state-of-art methods.
This paper presented a new method based on part matching to improve the accuracy of person re-identification. First of all,a person DPM which carried the information of person structure was used to segment the human body into parts with semantic meanings,and these parts were used as basic units for re-identification. Secondly,the deep features of these body parts were extracted by deep neural network model. Thirdly,each body part of the testing person was compared with the corresponding body part of the target person based on the deep feature and cosine distance. Finally,the matching results from all the body parts were fused to make the final decision. Experimental results show that this method is more robust and it outperforms most state-of-art methods.
引文
[1]Ying Shan,Sawhney H S,Kumar R.Unsupervised learning of discriminative edge measures for vehicle matching between non-overlapping cameras[J].IEEE Trans on Pattern Analysis and Machine Intelligence,2008,30:700-711.
[2]Zheng Weishi,Gong Shaogang,Xiang Tao.Person re-identification by probabilistic relative distance comparison[C]//Proc of IEEE Conference on Computer Vision and Pattern Recognition.2011:649-656.
[3]Li Wei,Wang Xiaogang.Locally aligned feature transforms across views[C]//Proc of IEEE Conference on Computer Vision and Pattern Recognition.2013:3594-3601.
[4]Alahi A,Vandergheynst P,Bierlaire M,et al.Cascade of descriptors to detect and track objects across any network of cameras[J].Computer Vision and Image Understanding,2010,114(6):624-640.
[5]贲晛烨,徐森,王科俊.行人步态的特征表达及识别综述[J].模式识别与人工智能,2012,25(1):71-81.
[6]范彩霞,朱虹,蔺广逢.多特征融合的人体目标再识别[J].中国图象图形学报,2013,18(6):711-717.
[7]曾明勇,吴泽民,田畅,等.基于外观统计特征融合的人体目标再识别[J].电子与信息学报,2014,36(8):1844-1851.
[8]Engel C,Baumgartner P,Holzmann M,et al.Person re-identification by support vector ranking[C]//Proc of IEEE Conference on British Machine Vision Conference.2010:1-11.
[9]Gray D,Tao Hai.Viewpoint invariant pedestrian recognition with an ensemble of localized features[C]//Proc of IEEE European Conference on Computer Vision.2008:262-275.
[10]Yi Dong,Lei Zhen,Liao Shengcai,et al.Deep metric learning for person re-identification[C]//Proc of International Conference on Pattern Recognition.2014:34-39.
[11]Xiong Fei,Gou Mengran,Camps O,et al.Person re-identification using kernel-based metric learning methods[C]//Proc of IEEE European Conference on Computer Vision.2014:1-16.
[12]Kstinger M,Hirzer M,Wohlhart P.Large scale metric learning from equivalence constraints[C]//Proc of IEEE Conference on Computer Vision and Pattern Recognition.2012:2288-2295.
[13]Bazzani L,Cristani M,Murino V.Symmetry-driven accumulation of local features for human characterization and re-identification[J].Computer Vision and Image Understanding,2013,117(2):130-144.
[14]Zhao Rui,Ouyang Wanli,Wang Xiaogang.Unsupervised salience learning for person re-identification[C]//Proc of IEEE Conference on Computer Vision and Pattern Recognition.2013:3586-3593.
[15]Felzenszwalb P F,Girshick R B,Mc Allester D.Object detection with discriminatively trained part based models[J].IEEE Trans on Pattern Analysis&Machine Intelligence,2010,32(9):1627-1645.
[16]Felzenszwalb P F,Huttenlocher D P.Pictorial structures for object recognition[J].International Journal of Computer Vision,2005,61(1):55-79.
[17]He Kaiming,Zhang Xiangyu,Ren Shaoqing,et al.Spatial pyramid pooling in deep convolutional networks for visual recognition[J].IEEE Trans on Pattern Analysis&Machine Intelligence,2015,37(9):1904-1916
[18]Everingham M,Van Gool L,Williams C K I.The PASCAL visual object classes(VOC)challenge[J].International Journal of Computer Vision,2010,88(2):303-338.
[19]Gray D,Brennan S,Tao Hai.Evaluating appearance models for recognition reacquisition,and tracking[C]//Proc of IEEE International Workshop on Performance Evaluation of Tracking and Surveillance.2007.
[20]Dikmen M,Akbas E,Huang T S,et al.Pedestrian recognition with a learned metric[C]//Lecture Notes in Computer Science,vol 6495.2011:501-512.
[21]Mignon A,Jurie F.PCCA:a new approach for distance learning from sparse pairwise constraints[C]//Proc of IEEE Conference on Computer Vision and Pattern Recognition.2012:2666-2672.
[22]Hirzer M,Roth P M,Kostinger M,et al.Relaxed pairwise learned metric for person re-identification[C]//Proc of IEEE European Conference on Computer Vision.2012:780-793.
[2]Zheng Weishi,Gong Shaogang,Xiang Tao.Person re-identification by probabilistic relative distance comparison[C]//Proc of IEEE Conference on Computer Vision and Pattern Recognition.2011:649-656.
[3]Li Wei,Wang Xiaogang.Locally aligned feature transforms across views[C]//Proc of IEEE Conference on Computer Vision and Pattern Recognition.2013:3594-3601.
[4]Alahi A,Vandergheynst P,Bierlaire M,et al.Cascade of descriptors to detect and track objects across any network of cameras[J].Computer Vision and Image Understanding,2010,114(6):624-640.
[5]贲晛烨,徐森,王科俊.行人步态的特征表达及识别综述[J].模式识别与人工智能,2012,25(1):71-81.
[6]范彩霞,朱虹,蔺广逢.多特征融合的人体目标再识别[J].中国图象图形学报,2013,18(6):711-717.
[7]曾明勇,吴泽民,田畅,等.基于外观统计特征融合的人体目标再识别[J].电子与信息学报,2014,36(8):1844-1851.
[8]Engel C,Baumgartner P,Holzmann M,et al.Person re-identification by support vector ranking[C]//Proc of IEEE Conference on British Machine Vision Conference.2010:1-11.
[9]Gray D,Tao Hai.Viewpoint invariant pedestrian recognition with an ensemble of localized features[C]//Proc of IEEE European Conference on Computer Vision.2008:262-275.
[10]Yi Dong,Lei Zhen,Liao Shengcai,et al.Deep metric learning for person re-identification[C]//Proc of International Conference on Pattern Recognition.2014:34-39.
[11]Xiong Fei,Gou Mengran,Camps O,et al.Person re-identification using kernel-based metric learning methods[C]//Proc of IEEE European Conference on Computer Vision.2014:1-16.
[12]Kstinger M,Hirzer M,Wohlhart P.Large scale metric learning from equivalence constraints[C]//Proc of IEEE Conference on Computer Vision and Pattern Recognition.2012:2288-2295.
[13]Bazzani L,Cristani M,Murino V.Symmetry-driven accumulation of local features for human characterization and re-identification[J].Computer Vision and Image Understanding,2013,117(2):130-144.
[14]Zhao Rui,Ouyang Wanli,Wang Xiaogang.Unsupervised salience learning for person re-identification[C]//Proc of IEEE Conference on Computer Vision and Pattern Recognition.2013:3586-3593.
[15]Felzenszwalb P F,Girshick R B,Mc Allester D.Object detection with discriminatively trained part based models[J].IEEE Trans on Pattern Analysis&Machine Intelligence,2010,32(9):1627-1645.
[16]Felzenszwalb P F,Huttenlocher D P.Pictorial structures for object recognition[J].International Journal of Computer Vision,2005,61(1):55-79.
[17]He Kaiming,Zhang Xiangyu,Ren Shaoqing,et al.Spatial pyramid pooling in deep convolutional networks for visual recognition[J].IEEE Trans on Pattern Analysis&Machine Intelligence,2015,37(9):1904-1916
[18]Everingham M,Van Gool L,Williams C K I.The PASCAL visual object classes(VOC)challenge[J].International Journal of Computer Vision,2010,88(2):303-338.
[19]Gray D,Brennan S,Tao Hai.Evaluating appearance models for recognition reacquisition,and tracking[C]//Proc of IEEE International Workshop on Performance Evaluation of Tracking and Surveillance.2007.
[20]Dikmen M,Akbas E,Huang T S,et al.Pedestrian recognition with a learned metric[C]//Lecture Notes in Computer Science,vol 6495.2011:501-512.
[21]Mignon A,Jurie F.PCCA:a new approach for distance learning from sparse pairwise constraints[C]//Proc of IEEE Conference on Computer Vision and Pattern Recognition.2012:2666-2672.
[22]Hirzer M,Roth P M,Kostinger M,et al.Relaxed pairwise learned metric for person re-identification[C]//Proc of IEEE European Conference on Computer Vision.2012:780-793.