基于多项式拟合的模特走秀动作分类方法
|
摘要
为自动、准确地对模特走秀动作进行评价,提出了一种基于多项式拟合的动作分类方法。该方法首先利用基于局部亲和域的方法进行人体关节点检测,同时为消除相机视角和个体体型差异性,将检测到的关节点通过普氏分析进行数据校准;其次将人体关节点分为脊柱、上肢和下肢三部分,分别从水平和垂直方向进行多项式拟合;再次对得到的多项式系数进行数据降维;最后将降维后的多项式系数作为动作评价的特征,利用SVM分类器实现模特走秀动作分类。实验结果表明:该方法评价准确率为71.9%,初步实现对人体动作的定性评价。该方法为模特走秀的动作评估提供了一种解决方案。
In order to evaluate catwalk show actions in an automatic and accurate way, an action classification method based on polynomial fitting is proposed in this paper. Firstly, the method uses the Part Affinity Fields to detect the human joint point. Data calibration of detected joint points was conducted by Prussian analysis to eliminate the difference between camera perspective and individual body type. Then, the detected human joint points were divided into three parts: spine, upper limb and lower limbs. Polynomial fitting of the data of the joint points was implemented respectively from the horizontal and vertical directions. Data dimensionality reduction was carried out again for polynomial coefficients obtained. Finally, the polynomial coefficient after reducing the dimension was regarded as the feature of the action evaluation, and SVM classifier was used to achieve classification of the model catwalk show actions. The experimental results showed that the evaluation accuracy of this method was 71.9%, and the qualitative evaluation of human body's action was preliminarily realized. It provides an effective solution for the catwalk action assessment of the models.
In order to evaluate catwalk show actions in an automatic and accurate way, an action classification method based on polynomial fitting is proposed in this paper. Firstly, the method uses the Part Affinity Fields to detect the human joint point. Data calibration of detected joint points was conducted by Prussian analysis to eliminate the difference between camera perspective and individual body type. Then, the detected human joint points were divided into three parts: spine, upper limb and lower limbs. Polynomial fitting of the data of the joint points was implemented respectively from the horizontal and vertical directions. Data dimensionality reduction was carried out again for polynomial coefficients obtained. Finally, the polynomial coefficient after reducing the dimension was regarded as the feature of the action evaluation, and SVM classifier was used to achieve classification of the model catwalk show actions. The experimental results showed that the evaluation accuracy of this method was 71.9%, and the qualitative evaluation of human body's action was preliminarily realized. It provides an effective solution for the catwalk action assessment of the models.
引文
[1] Hofmann M, Gavrila D M. Multi-view 3D human pose estimation in complex environment[J]. International Journal of Computer Vision, 2012, 96(1): 103-124.
[2] Chan C K, Loh W P, Rahim I A. Human motion classification using 2D stick-model matching regression coefficients[J]. Applied Mathematics and Computation, 2016, 283(12): 70-89.
[3] 佟丽娜, 侯增广, 彭亮, 等. 基于多路 sEMG 时序分析的人体运动模式识别方法[J]. 自动化学报, 2014, 40(5): 810-821.
[4] Wolf A, Senesh M. Motion estimation using a statistical solid dynamic method[M]// Kecskeméthy A, Müller A. Computational Kinematics. Springer, Berlin, Heidelberg, 2009: 109-116.
[5] 刘磊, 杨鹏, 刘作军. 采用多核相关向量机的人体步态识别[J]. 浙江大学学报 (工学版), 2017, 51(3): 562-571.
[6] 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.
[7] Fischler M A, Elschlager R A. The representation and matching of pictorial structures[J]. IEEE Transactions on Computers, 1973, 100(1): 67-92.
[8] Dubout C, Fleuret F. Exact acceleration of linear object detectors[C]//European Conference on Computer Vision. Springer, Berlin, Heidelberg, 2012: 301-311.
[9] Pantic M, Rothkrantz L J M. Automatic analysis of facial expressions: The state of the art[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2000, 22(12): 1424-1445.
[10] Ouyang W, Chu X, Wang X. Multi-source deep learning for human pose estimation[C]// Computer Vision and Pattern Recognition. IEEE, 2014:2337-2344.
[11] Chen X, Yuille A. Articulated pose estimation by a graphical model with image dependent pairwise relations[EB/OL].(2014-11-04)[2018-06-11]. https://arxiv.org/abs/1407.3399.
[12] Jain A. Articulated people detection and pose estimation: Reshaping the future[C]// Computer Vision and Pattern Recognition. IEEE, 2012:3178-3185.
[13] Gkioxari G, Hariharan B, Girshick R, et al. Using k-poselets for detecting people and localizing their keypoints[C]// IEEE Conference on Computer Vision and Pattern Recognition. IEEE Computer Society, 2014:3582-3589.
[14] Cao Z, Simon T, Wei S E, et al. Realtime multi-person 2D pose estimation using Part Affinity Fields[C]// IEEE Conference on Computer Vision and Pattern Recognition. IEEE, 2017:1302-1310.
[15] 毛红保, 张凤鸣, 冯卉. 基于奇异值分解的飞行动作评价方法研究[J]. 计算机工程与应用, 2008, 44(32):240-242.
[16] Ren S, Cao X, Wei Y, et al. Face alignment at 3000 fps via regressing local binary features[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. IEEE,2014: 1685-1692.
[17] 杜克勤. Procrustes问题的若干研究[D].杭州:浙江大学,2005:7-22.
[18] Wei S E, Ramakrishna V, Kanade T, et al. Convolutional pose machines[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. IEEE,2016: 4724-4732.
[19] Güdükbay U, Demir I, Dedeo. Motion capture and human pose reconstruction from a single-view video sequence[J]. Digital Signal Processing, 2013, 23(5): 1441-1450.
[20] 孟强, 王佩贤, 刘志朋. 基于最小二乘法的高速道岔精测技术研究[J]. 测绘与空间地理信息, 2018, 41 (2):186-189.
[21] 苑玮琦, 曲晓峰, 柯丽, 等. 主成分分析重建误差掌纹识别方法[J]. 光学学报, 2008, 28(10): 1903-1909.
[2] Chan C K, Loh W P, Rahim I A. Human motion classification using 2D stick-model matching regression coefficients[J]. Applied Mathematics and Computation, 2016, 283(12): 70-89.
[3] 佟丽娜, 侯增广, 彭亮, 等. 基于多路 sEMG 时序分析的人体运动模式识别方法[J]. 自动化学报, 2014, 40(5): 810-821.
[4] Wolf A, Senesh M. Motion estimation using a statistical solid dynamic method[M]// Kecskeméthy A, Müller A. Computational Kinematics. Springer, Berlin, Heidelberg, 2009: 109-116.
[5] 刘磊, 杨鹏, 刘作军. 采用多核相关向量机的人体步态识别[J]. 浙江大学学报 (工学版), 2017, 51(3): 562-571.
[6] 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.
[7] Fischler M A, Elschlager R A. The representation and matching of pictorial structures[J]. IEEE Transactions on Computers, 1973, 100(1): 67-92.
[8] Dubout C, Fleuret F. Exact acceleration of linear object detectors[C]//European Conference on Computer Vision. Springer, Berlin, Heidelberg, 2012: 301-311.
[9] Pantic M, Rothkrantz L J M. Automatic analysis of facial expressions: The state of the art[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2000, 22(12): 1424-1445.
[10] Ouyang W, Chu X, Wang X. Multi-source deep learning for human pose estimation[C]// Computer Vision and Pattern Recognition. IEEE, 2014:2337-2344.
[11] Chen X, Yuille A. Articulated pose estimation by a graphical model with image dependent pairwise relations[EB/OL].(2014-11-04)[2018-06-11]. https://arxiv.org/abs/1407.3399.
[12] Jain A. Articulated people detection and pose estimation: Reshaping the future[C]// Computer Vision and Pattern Recognition. IEEE, 2012:3178-3185.
[13] Gkioxari G, Hariharan B, Girshick R, et al. Using k-poselets for detecting people and localizing their keypoints[C]// IEEE Conference on Computer Vision and Pattern Recognition. IEEE Computer Society, 2014:3582-3589.
[14] Cao Z, Simon T, Wei S E, et al. Realtime multi-person 2D pose estimation using Part Affinity Fields[C]// IEEE Conference on Computer Vision and Pattern Recognition. IEEE, 2017:1302-1310.
[15] 毛红保, 张凤鸣, 冯卉. 基于奇异值分解的飞行动作评价方法研究[J]. 计算机工程与应用, 2008, 44(32):240-242.
[16] Ren S, Cao X, Wei Y, et al. Face alignment at 3000 fps via regressing local binary features[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. IEEE,2014: 1685-1692.
[17] 杜克勤. Procrustes问题的若干研究[D].杭州:浙江大学,2005:7-22.
[18] Wei S E, Ramakrishna V, Kanade T, et al. Convolutional pose machines[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. IEEE,2016: 4724-4732.
[19] Güdükbay U, Demir I, Dedeo. Motion capture and human pose reconstruction from a single-view video sequence[J]. Digital Signal Processing, 2013, 23(5): 1441-1450.
[20] 孟强, 王佩贤, 刘志朋. 基于最小二乘法的高速道岔精测技术研究[J]. 测绘与空间地理信息, 2018, 41 (2):186-189.
[21] 苑玮琦, 曲晓峰, 柯丽, 等. 主成分分析重建误差掌纹识别方法[J]. 光学学报, 2008, 28(10): 1903-1909.