Spatial Pyramid Pooling in Deep Convolutional Networks for Visual Recognition

被引:4452
作者
He, Kaiming [1 ]
Zhang, Xiangyu [2 ]
Ren, Shaoqing [3 ]
Sun, Jian [1 ]
机构
[1] Visual Comp Grp, Microsoft Res, Beijing 100080, Peoples R China
[2] Xi An Jiao Tong Univ, Dept Elect Engn, Xian 710049, Peoples R China
[3] Univ Sci & Technol China, Dept Elect Sci & Technol, Hefei 230026, Peoples R China
关键词
Convolutional neural networks; spatial pyramid pooling; image classification; object detection;
D O I
10.1109/TPAMI.2015.2389824
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Existing deep convolutional neural networks (CNNs) require a fixed-size (e.g., 224x224) input image. This requirement is "artificial" and may reduce the recognition accuracy for the images or sub-images of an arbitrary size/scale. In this work, we equip the networks with another pooling strategy, "spatial pyramid pooling", to eliminate the above requirement. The new network structure, called SPP-net, can generate a fixed-length representation regardless of image size/scale. Pyramid pooling is also robust to object deformations. With these advantages, SPP-net should in general improve all CNN-based image classification methods. On the ImageNet 2012 dataset, we demonstrate that SPP-net boosts the accuracy of a variety of CNN architectures despite their different designs. On the Pascal VOC 2007 and Caltech101 datasets, SPP-net achieves state-of-the-art classification results using a single full-image representation and no fine-tuning. The power of SPP-net is also significant in object detection. Using SPP-net, we compute the feature maps from the entire image only once, and then pool features in arbitrary regions (sub-images) to generate fixed-length representations for training the detectors. This method avoids repeatedly computing the convolutional features. In processing test images, our method is 24-102x faster than the R-CNN method, while achieving better or comparable accuracy on Pascal VOC 2007. In ImageNet Large Scale Visual Recognition Challenge (ILSVRC) 2014, our methods rank #2 in object detection and #3 in image classification among all 38 teams. This manuscript also introduces the improvement made for this competition.
引用
收藏
页码:1904 / 1916
页数:13
相关论文
共 39 条
  • [1] [Anonymous], 2014, ARXIV14031840
  • [2] [Anonymous], 2013, Caffe: An open source convolutional architecture for fast feature embedding
  • [3] [Anonymous], ARXIV14053531
  • [4] LIBSVM: A Library for Support Vector Machines
    Chang, Chih-Chung
    Lin, Chih-Jen
    [J]. ACM TRANSACTIONS ON INTELLIGENT SYSTEMS AND TECHNOLOGY, 2011, 2 (03)
  • [5] The devil is in the details: an evaluation of recent feature encoding methods
    Chatfield, Ken
    Lempitsky, Victor
    Vedaldi, Andrea
    Zisserman, Andrew
    [J]. PROCEEDINGS OF THE BRITISH MACHINE VISION CONFERENCE 2011, 2011,
  • [6] Coates A., 2011, International Conference on Machine Learning, P921
  • [7] Histograms of oriented gradients for human detection
    Dalal, N
    Triggs, B
    [J]. 2005 IEEE COMPUTER SOCIETY CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, VOL 1, PROCEEDINGS, 2005, : 886 - 893
  • [8] Deng J, 2009, PROC CVPR IEEE, P248, DOI 10.1109/CVPRW.2009.5206848
  • [9] Donahue J., 2013, CoRR
  • [10] The PASCAL Visual Object Classes Challenge: A Retrospective
    Everingham, Mark
    Eslami, S. M. Ali
    Van Gool, Luc
    Williams, Christopher K. I.
    Winn, John
    Zisserman, Andrew
    [J]. INTERNATIONAL JOURNAL OF COMPUTER VISION, 2015, 111 (01) : 98 - 136