Image classification method based on local receptive field extended S-Mobilenet model

Hongmei Liu; Chuanwu Tan

doi:10.54097/hset.v7i.994

Authors

Hongmei Liu
Chuanwu Tan

DOI:

https://doi.org/10.54097/hset.v7i.994

Keywords:

Image classification, Deep neural network, MobileNet, Cavity convolution, S-MobileNet

Abstract

Aiming at the problem that the lightweight deep neural network mobilenet will reduce the classification accuracy, the hollow convolution kernel is introduced into a convolution layer of mobilenet model, and a S-Mobilenet model based on local receptive field expansion is proposed. The model is divided into three structures according to the location of the hole convolution kernel. Without increasing the number of parameters, it can expand the local receptive field of the convolution kernel and improve the classification accuracy. The experiment was carried out on caltech-101 data set, caltech-256 data set and animal classification database of Tubingen University. The results show that S-Mobilenet model can obtain better classification accuracy than Mobilenet, which can be improved by 2% at most.

Downloads

Download data is not yet available.

References

Krizhevsky A, Sutskever I, Hinton G E. ImageNet classification with deep convolutional neural networks [J]. Communications of the ACM, 2017, 60(6): 84-90.

Wang Naiyan, Yeung D Y. Learning a deep compact image representa- tion for visual tracking [C]/ / Proc of the 26th International Confe- rence on Neural Information Processing Systems. [S.l.]: Curran As- sociates Inc., 2013:809-817.

Wan Ji,Wang Dayong, Hoi S C H, et al. Deep learning for content- based image retrieval: a comprehensive study [C]/ / Proc of the 22nd ACMInternational Conference on Multimedia.New York:ACMPress, 2014: 157-166.

Chen L C, Papandreou G, Kokkinos I, et al. DeepLab: semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected CRFS [J]. IEEE Trans on Pattern Analysis and Machine Intelligence, 2018, 40(4): 834-848.

Iandola F N, Han S, Moskewicz MW, et al.Heatmap: AlexNet-level accuracy with 50x fewer parameters and ＜0.5 MB model size [EB/ OL]. (2016-11-04). https: //arxiv.org/abs/1602. 07360.

Livny Y, Yan Feilong, Chen Baoquan. Automatic reconstruction of tree skeletal structures from point clouds [J]. Association for Computing Machinery, 2010: article No.15 1.

Howard A G, Zhu Menglong, Chen Bo, et al.MobileNets: efficient con- volutional neural networks for mobile vision applications [EB/OL]. (2017-04-17). https: //arxiv.org/abs/1704. 04861.

Ioffe S, Szegedy C. Batch normalization: accelerating deep network training by reducing internal covariate shift [C]/ / Proc of the 32nd In- ternational Conference on International Conference on Machine Lear- ning.2015: 448-456.