期刊文献+

融合多姿势估计特征的动作识别 被引量:5

Fusing multiple pose estimations for still image action recognition
原文传递
导出
摘要 目的为了提高静态图像在遮挡等复杂情况下的动作识别效果和鲁棒性,提出融合多种姿势估计得到的特征信息进行动作识别的方法。方法利用已得到的多个动作模型对任意一幅图像进行姿势估计,得到图像的多组姿势特征信息,每组特征信息包括关键点信息和姿势评分。将训练集中各个动作下所有图像的区分性关键点提取出来,并计算每一幅图像中区分性关键点之间的相对距离,一个动作所有图像的特征信息共同构成该动作的模板信息。测试图像在多个动作模型下进行姿势估计,得到多组姿势特征,从每组姿势特征中提取与对应模板一致的特征信息,将提取的多组姿势特征信息分别与对应的模板进行匹配,并通过姿势评分对匹配值优化,根据最终匹配值进行动作分类。结果在两个数据集上,本文方法与5种比较流行的动作识别方法进行比较,获得了较好的平均准确率,在数据集PASCAL VOC 2011-val上较其他一些最新的经典方法平均准确率至少提高近2%。在数据集Stanford 40 actions上,较其他一些最新的经典方法平均准确率至少提高近6%。结论本文方法融合了多个姿势特征,并且能够获取关键部位的遮挡信息,所以能较好应对遮挡等复杂环境情况,具有较高的平均识别准确率。 Objective To adapt to occlusion or other complex situations, an action recognition method is proposed which fuses multiple pose estimation features. Method Multiple pose features will be obtained using muhiple action models. Each pose feature information includes key point positions and pose scores. Distinguishing key pointsare extracted from all train images and computing relative distances between point pairs. An action template is built using all features of the train images of the action. Multiple feature information consistent with multiple action templates are extracted from each test image from multiple pose features. Multiple features of the test image are-matched with the corresponding action template and then matched values are optimized using pose scores. Result The experimental results have shown that the average accuracy of the proposed method is approximately 2% better than some other state-of-the-art methods on VOC 2011-val set, and is approximately 6% better than some other state-of-the-art methods on Stanford 40 actions set. Conclusion By fusing muhiple pose features, the proposed method can adapt to occlusion and other complex situations and improve average recognition accuracy.
出处 《中国图象图形学报》 CSCD 北大核心 2015年第11期1462-1472,共11页 Journal of Image and Graphics
基金 国家自然科学基金项目(61105042 61462035) 国家重点基础研究发展计划(973)基金项目(2010CB327900) 江西省青年科学家(井冈之星)培养对象计划资助~~
关键词 动作识别 多姿势估计 模板匹配 遮挡 action recognition multiple pose estimations template matching occlusion
  • 相关文献

参考文献16

  • 1Yang W,Wang Y,Mori G.Recognizing human actions from still images with latent poses[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.San Francisco,CA:IEEE,2010:2030-2037.
  • 2Yang Y,Ramanan D.Articulated pose estimation with flexi- blemixtures-of-parts[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Colorado Springs,CO:IEEE,2011:1385-1392.
  • 3Yao B,Fei-Fei L.Action recognition with exemplar based 2.5 D graph matching[C]//Proceedings of the 12th European Confer- ence on Computer Vision.Berlin:Springer,2012:173-186.
  • 4Khan F S,Weijer vandej,Rao M A,et al.Semantic pyramids for gender and action recognition[J].IEEE Transactions on Im- age Processing,2014,23(8):3633-3645.
  • 5Desai C,Ramanan D.Detecting actions,poses,and objects with relational phraselets[C]//Proceedings of the 12th European Conference on Computer Vision.Berlin:Springer,2012:158-172.
  • 6Sharma G,Jurie F,Schmid C.Expanded parts model for human attribute and action recognition in still images[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Portland,OR:IEEE,2013:652-659.
  • 7Sadeghi M A,Farhadi A.Recognition using visual phrases[C]// Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Colorado Springs,CO:IEEE,2011:1745-1752.
  • 8Felzenszwalb P F,Huttenlocher D P.Pictorial structures for ob- ject recognition[J].International Journal of Computer Vision,2005,61(1):55-79.
  • 9Khan F S,Anwer R M,Weijer van de J,et al.Coloring action recognition in still images[J].International Journal of Computer Vision,2013,105(3):205-221.
  • 10Delaitre V,Laptev I,Sivic J.Recognizing human actions in still images:a study of bag-of-features and part-based representations [C]//Proceedings of the 21st British Machine Vision Confer- ence.Aberystwyth:BMVA,2010:97.1-97.11.

二级参考文献47

  • 1Efros A A, Berg A C, Mori G, et al. Recognizing action at a dis- tance[ C ]//Proceedings of the 9th IEEE International Conference on Computer Vision. Nice, France: IEEE, 2003:726-733.
  • 2Yilmaz A, Shah M. Matching actions in presence of camera mo- tion[ J]. Computer Vision and Image Understanding, 2006, 104(2-3) :221-231.
  • 3Dollar P, Rabaud V, Cottrell G, et al. Behavior recognition via sparse spatio-temporal features [ C ]//Proceedings of the 2nd Joint 1EEE International Workshop on :isual Surveillance and Perform- ance Evaluation of Tracking and Surveillance. Beijing, China: IEEE, 2005:65-72.
  • 4Laptev L On space-time interest points [ J ]. International Journal of Computer Vision, 2005, 64 (2) : 107-123.
  • 5Bregonzio M, Gong S, Xiang T. Recognising action as clouds of space-time interest points[ C ]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Florida, USA: IEEE, 2009 : 1948-1955.
  • 6Niebles J C, Wang H, Li F F. Unsupervised learning of human action categories using spatial-temporal words [ J ]. International Journal of Computer Vision, 2008,79(3): 299-318.
  • 7Laptev I, Marszalek M, Schmid C, et al. Learning realistic hu- man actions from movies [ C ]// Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Anchorage, AK, USA : IEEE, 2008 : 1-8.
  • 8Klaser A, M Marszalek, C Schmid. A spatio-temporal descriptor based on 3D-gradients [ C ]//Proceedings of British Machine Vi- sion Conference. Leeds, England : University of Leeds, 2008 : 955-1004.
  • 9Gao Z, Chen M, Hauptmann A, et at. Comparing evaluation protocols on the KTH dataset [ C ]//Proceedings of the 1 st Inter- national Conference on Human Behavior Understanding. Istan- bul, Turkey : Springer-Verlag, 2010:88-100.
  • 10Nater F, Grabner H, Gool L Van. Exploiting simple hierarchies for unsupervised human behavior analysis [ C ]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. San Francisco, California, USA: IEEE, 2010 : 2014-2021.

共引文献36

同被引文献36

引证文献5

二级引证文献41

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部