结合快速特征金字塔计算的可变形部件模型
|
摘要
针对可变形部件模型中特征金字塔需要精细计算,从而导致计算速度较慢这一问题,提出一种结合快速特征金字塔的可变形部件模型对象检测算法.首先在特征金字塔中根据间隔选取若干个基准尺度进行精确计算,然后依据多尺度特征中的幂指定律,近似计算得到多分辨率的图像特征.算法采用尺度上稀疏采样的特征金字塔来外推精细采样的特征金字塔,然后采用预先训练好的模板进行类别检测,得出检测结果.在PASCAL VOC2007以及INRIA数据集上的实验结果表明,该算法可以明显加速模型中特征计算的速度,而检测精度仅略有下降.
To solve the problem that feature pyramids of deformable part models being computed explicitly lead to slowcomputation,the paper proposed an object detection algorithm based on deformable part models with fast computation of feature pyramids. Firstly we select and compute explicitly several benchmark scales based on scale intervals on feature pyramids,then approximately compute multi-resolution image features according to the multi-scale power law. We utilize sparsely-sampled feature pyramids on the scale to approximate the finely-sampled feature pyramids,and then use pre-trained templates to detect the category to obtain final detection results. The experimental results on Pascal VOC 2007 dataset and INRIA dataset showthat the algorithm in the paper apparently accelerates the speed of feature computation in the model with negligible loss in detection accuracy.
To solve the problem that feature pyramids of deformable part models being computed explicitly lead to slowcomputation,the paper proposed an object detection algorithm based on deformable part models with fast computation of feature pyramids. Firstly we select and compute explicitly several benchmark scales based on scale intervals on feature pyramids,then approximately compute multi-resolution image features according to the multi-scale power law. We utilize sparsely-sampled feature pyramids on the scale to approximate the finely-sampled feature pyramids,and then use pre-trained templates to detect the category to obtain final detection results. The experimental results on Pascal VOC 2007 dataset and INRIA dataset showthat the algorithm in the paper apparently accelerates the speed of feature computation in the model with negligible loss in detection accuracy.
引文
[1]Xu J,Vazquez D,Lopez A M,et al.Learning a part-based pedestrian detector in a virtual w orld[J].IEEE Transactions on Intelligent Transportation Systems,2014,15(5):2121-2131.
[2]Yin Xue-cong.Study on face detection based on deformable part model[D].Xi'ian Xi'ian University,2012.
[3]Artan Y,Paul P,Perronin F,et al.Comparison of face detection and image classification for detecting front seat passengers in vehicles[C].Proceedings of IEEE Winter Conference on Applications of Computer Vision,2014:1006-1012.
[4]Huang Kai-qi,Ren Wei-qiang,Tan Tie-niu.A review on image object classification and detection[J].Chinese Journal of Computers,2014,37(6):1225-1240.
[5]Dalal N,Triggs B.Histograms of oriented gradients for human detection[C].Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition,2005:886-893.
[6]Felzenszwalb P F,Girshick R B,Mcallester D,et al.Object detection w ith discriminatively trained part-based models[J].Pattern Analysis&M achine Intelligence,2010,32(9):1627-1645.
[7]Felzenszwalb P F,Girshick R B,Mc Allester D.Cascade object detection w ith deformable part models[C].Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition,2010:2241-2248.
[8]Pedersoli M,Vedaldi A,Gonzalez J.A coarse-to-fine approach for fast deformable object detection[C].Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2011:1353-1360.
[9]Kokkinos I.Bounding part scores for rapid detection with deformable part models[M].Computer Vision-ECCV 2012,Workshops and Demonstrations Springer,2012:41-50.
[10]Ramanan D,Pirsiavash H.Steerable part models[C].Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2012:3226-3233.
[11]Song H O,Zickler S,Althoff T,et al.Sparselet models for efficient multiclass object detection[C].Proceedings of the 12th European Conference on Computer Vision,2012:802-815.
[12]Dean T,Ruzon M,Segal M,et al.Fast,accurate detection of 100,000 object classes on a single machine[C].Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2013:1814-1821.
[13]Yang Yang,Li Shan-ping.Fast object detection with deformable part models and segment locations'hint[J].Acta Automatica Sinica,2012,38(4):540-548.
[14]Dollar P,Appel R,Belongie S,et al.Fast feature pyramids for object detection[J].Pattern Analysis&M achine Intelligence,2014,36(8):1532-1545.
[15]Felzenszwalb P F,Huttenlocher D.Pictorial structures for object recognition[J].International Journal of Computer Vision,2005,61(6):55-79.
[16]Ruderman D L,Bialek W.Statistics of natural images:scaling in the w oods[J].Physical Review Letters,1994,73(6):814-817.
[17]Ruderman D L.The statistics of natural images[J].Network Computation in Neural Systems,2009,5(4):517-548.
[18]Dollár P,Belongie S,Perona P.The fastest pedestrian detector in the w est[C].Proceedings of the IEEE Conference on British M achine Vision Conference,2010:1-11.
[2]尹雪聪.基于可变形部件模型的人脸检测方法研究[D].西安:西安电子科技大学,2012.
[4]黄凯奇,任伟强,谭铁牛.图像物体分类与检测算法综述[J].计算机学报,2014,37(6):1225-1240.
[13]杨扬,李善平.分割位置提示的可变形部件模型快速目标检测[J].自动化学报,2012,38(4):540-548.
[2]Yin Xue-cong.Study on face detection based on deformable part model[D].Xi'ian Xi'ian University,2012.
[3]Artan Y,Paul P,Perronin F,et al.Comparison of face detection and image classification for detecting front seat passengers in vehicles[C].Proceedings of IEEE Winter Conference on Applications of Computer Vision,2014:1006-1012.
[4]Huang Kai-qi,Ren Wei-qiang,Tan Tie-niu.A review on image object classification and detection[J].Chinese Journal of Computers,2014,37(6):1225-1240.
[5]Dalal N,Triggs B.Histograms of oriented gradients for human detection[C].Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition,2005:886-893.
[6]Felzenszwalb P F,Girshick R B,Mcallester D,et al.Object detection w ith discriminatively trained part-based models[J].Pattern Analysis&M achine Intelligence,2010,32(9):1627-1645.
[7]Felzenszwalb P F,Girshick R B,Mc Allester D.Cascade object detection w ith deformable part models[C].Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition,2010:2241-2248.
[8]Pedersoli M,Vedaldi A,Gonzalez J.A coarse-to-fine approach for fast deformable object detection[C].Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2011:1353-1360.
[9]Kokkinos I.Bounding part scores for rapid detection with deformable part models[M].Computer Vision-ECCV 2012,Workshops and Demonstrations Springer,2012:41-50.
[10]Ramanan D,Pirsiavash H.Steerable part models[C].Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2012:3226-3233.
[11]Song H O,Zickler S,Althoff T,et al.Sparselet models for efficient multiclass object detection[C].Proceedings of the 12th European Conference on Computer Vision,2012:802-815.
[12]Dean T,Ruzon M,Segal M,et al.Fast,accurate detection of 100,000 object classes on a single machine[C].Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2013:1814-1821.
[13]Yang Yang,Li Shan-ping.Fast object detection with deformable part models and segment locations'hint[J].Acta Automatica Sinica,2012,38(4):540-548.
[14]Dollar P,Appel R,Belongie S,et al.Fast feature pyramids for object detection[J].Pattern Analysis&M achine Intelligence,2014,36(8):1532-1545.
[15]Felzenszwalb P F,Huttenlocher D.Pictorial structures for object recognition[J].International Journal of Computer Vision,2005,61(6):55-79.
[16]Ruderman D L,Bialek W.Statistics of natural images:scaling in the w oods[J].Physical Review Letters,1994,73(6):814-817.
[17]Ruderman D L.The statistics of natural images[J].Network Computation in Neural Systems,2009,5(4):517-548.
[18]Dollár P,Belongie S,Perona P.The fastest pedestrian detector in the w est[C].Proceedings of the IEEE Conference on British M achine Vision Conference,2010:1-11.
[2]尹雪聪.基于可变形部件模型的人脸检测方法研究[D].西安:西安电子科技大学,2012.
[4]黄凯奇,任伟强,谭铁牛.图像物体分类与检测算法综述[J].计算机学报,2014,37(6):1225-1240.
[13]杨扬,李善平.分割位置提示的可变形部件模型快速目标检测[J].自动化学报,2012,38(4):540-548.