DOG: A new background removal for object recognition from images

Abstract For image classification, convolutional neural networks (CNNs) have the advantage of being able to convolve directly with image pixels and extract image features from image pixels. This approach is closer to the treatment of the human brain's visual system. However, up to now, it is impossible to achieve 100% classification accuracy regardless of any kind of convolutional neural network models. At the same time, we also find that sometimes the background of images affects the recognition of neural networks, and removing the background of images can improve their performance of object recognition. Therefore, we design a model named DOG based on CNN to remove the background of images (such as selfies, animals, flowers, etc.) for improving the performance of object recognition. Because of the scarce of samples, we further integrate DOG with DCGAN to further improve the performance of object recognition. Our experimental results show the effectiveness of DOG.

[1]  Yong Peng,et al.  Discriminative extreme learning machine with supervised sparsity preserving for image classification , 2017, Neurocomputing.

[2]  Lucia Maddalena,et al.  Extensive Benchmark and Survey of Modeling Methods for Scene Background Initialization , 2017, IEEE Transactions on Image Processing.

[3]  Marc Van Droogenbroeck,et al.  LaBGen-P: A pixel-level stationary background generation method based on LaBGen , 2016, 2016 23rd International Conference on Pattern Recognition (ICPR).

[4]  Andrew Zisserman,et al.  Very Deep Convolutional Networks for Large-Scale Image Recognition , 2014, ICLR.

[5]  Kunihiko Fukushima,et al.  Neocognitron: A Self-Organizing Neural Network Model for a Mechanism of Visual Pattern Recognition , 1982 .

[6]  Lawrence D. Jackel,et al.  Handwritten Digit Recognition with a Back-Propagation Network , 1989, NIPS.

[7]  Fahd Bouzaraa,et al.  CNN-based initial background estimation , 2016, 2016 23rd International Conference on Pattern Recognition (ICPR).

[8]  Yoshua Bengio,et al.  Gradient-based learning applied to document recognition , 1998, Proc. IEEE.

[9]  Tao Wang,et al.  End-to-end text recognition with convolutional neural networks , 2012, Proceedings of the 21st International Conference on Pattern Recognition (ICPR2012).

[10]  Rob Fergus,et al.  Visualizing and Understanding Convolutional Networks , 2013, ECCV.

[11]  Wei Fang,et al.  A Novel Convolution Neural Network for Background Segmentation Recognition , 2018, ICCCS.

[12]  Dumitru Erhan,et al.  Going deeper with convolutions , 2014, 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[13]  Bin Gu,et al.  A Solution Path Algorithm for General Parametric Quadratic Programming Problem , 2018, IEEE Transactions on Neural Networks and Learning Systems.

[14]  Bin Gu,et al.  A regularization path algorithm for support vector ordinal regression , 2018, Neural Networks.

[15]  Thomas Brox,et al.  U-Net: Convolutional Networks for Biomedical Image Segmentation , 2015, MICCAI.

[16]  Yoshua Bengio,et al.  The One Hundred Layers Tiramisu: Fully Convolutional DenseNets for Semantic Segmentation , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW).

[17]  Hecht-Nielsen.  Theory of the backpropagation neural network , 1989 .

[18]  Pietro Perona,et al.  Microsoft COCO: Common Objects in Context , 2014, ECCV.

[19]  D. Hubel,et al.  Receptive fields and functional architecture of monkey striate cortex , 1968, The Journal of physiology.

[20]  Kilian Q. Weinberger,et al.  Densely Connected Convolutional Networks , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[21]  Geoffrey E. Hinton,et al.  Rectified Linear Units Improve Restricted Boltzmann Machines , 2010, ICML.

[22]  Ian Goodfellow,et al.  Generative adversarial networks , 2020, Commun. ACM.

[23]  Soumith Chintala,et al.  Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks , 2015, ICLR.