针对目前的深度卷积神经网络(CNN)模型规模大、训练参数多、计算速度慢以及难以移植到移动端等问题,提出了一种深度可分离卷积结合3重注意机制模块(DSC-TAM)的视觉模型。首先,通过深度可分离卷积网络来减少模型参数,提高网络模型的计算速度;其次,引入3重注意机制模块提高网络的特征提取能力,改善网络性能。实验结果表明:该方法的识别率可达99.63%,模型规模降低了13%;与标准卷积神经网络视觉模型及其他方法比较,在保证识别精度的同时减少了网络模型的大小。
Abstract
To solve the problems of current deep convolutional neural network (CNN), such as large model size, many training parameters, slow computing speed, and difficulty in transplantation to mobile terminal, a visual model of depthwise separable convolution with triple attention module (DSC-TAM) is proposed. Firstly, the depthwise separable convolution is used to reduce the model parameters and improve the computing speed of the network model. Secondly, the triple attention mechanism module is introduced to improve the ability of feature extraction and network performance. The experimental results show that the recognition rate of this method is 99.63%, the model size is reduced by 13%. Compared with the standard convolutional neural network visual model and other methods, the recognition accuracy is guaranteed, and the size of the network model is reduced.
关键词
计量学 /
视觉模型 /
3重注意机制 /
深度可分离卷积 /
神经网络 /
目标识别
Key words
metrology /
visual model /
triple attention mechanism /
depthwise separable convolution /
neural network /
target recognition
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
［1］Szegedy C, Wei L, Jia V, et al. Going Deeper with Convolutions［C］// 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Boston, USA, 2015, 1-9.
［2］Simonyan K, Zisserman A. Very Deep Convolutional Networks for Large-scale Image Recognition［J］. Computer Science, 2014, arXiv:1409.1556.
［3］He K, Zhang X, Ren S, et al. Deep Residual Learning for Image Recognition［C］//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, USA, 2016, 770-778.
［4］张世辉,王红蕾,陈宇翔,等.基于深度学习利用特征图加权融合的目标检测方法［J］. 计量学报, 2020, 41(11):1344-1351.
Zhang S H, Wang H L,Chen Y X,et al. A Target Detection Method Based on Deep Learning and Weighted Fusion of Feature Images［J］.Acta Metrologica Sinica,2020,41(11):1344-1351.
［5］Zoph B, Vasudevan V, Shlens J, et al. Learning Transferable Architectures for Scalable Image Recognition［C］//2018 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Salt Lake City, USA, 2018, 8697-8710.
［6］Real E, Aggarwal A, Huang Y, et, al. Regularized Evolution for Image Classifier Architecture Search［J］. AAAI, 2019, 33(1): 4780-4789.
［7］Szegedy C, Ioffe S, Vanhoucke V, et al. Inception-ResNet and the Impact of Residual Connections on Learning［J］. Computer Science, 2016, 1602.07261.
［8］Han S, Mao H, Dally W, Deep Compression: Compressing Deep Neural Networks with Pruning［J］. Trained Quantization and Huffman Coding, Fiber, 2015, 56(4): 3-7.
［9］Itti L, Koch C, Niebur E. A Model of Saliency-based Visual Attention for Rapid Scene Analysis［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1998, 20(11): 1254-1259.
［10］Jaderberg M, Simonyan K, Zisserman A, et al. Spatial transformer networks［C］//Proceedings of the 28th International Conference on Neural Information Processing Systems-Volume 2. Montreal , Canada, 2015, 2017-2025..
［11］Hu J, Shen L and Sun G, Squeeze-and-Excitation Networks ［C］// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, USA, 2018, 7132-7141.
［12］Woo S, Park J, Lee J Y, et al. Cbam: Convolutional block attention module［C］//Proceedings of the European conference on computer vision (ECCV), Munich, Germang, 2018, 3-19.
［13］程淑红, 张仕军, 赵考鹏. 基于卷积神经网络的生物式水质监测方法［J］. 计量学报, 2019, 40(4): 721-727.
Cheng S H, Zhang S J, Zhao K P. Biologic Water Quality Monitoring Method Based on Convolutional Neural Network［J］. Acta Metrologica Sinica, 2019, 40(4): 721-727.
［14］程涛. 基于深度分离卷积神经网络的机器人花卉分拣系统［D］. 长沙: 湖南工业大学, 2019.
［15］唐永昊. 基于视觉注意模型的SAR图像目标检测算法研究［D］. 成都: 电子科技大学, 2020.
［16］Xin R, Zhang J, Shao Y. Complex Network Classification with Convolutional Neural Network［J］. Tsinghua Science and Technology, 2020, 25(4): 447-457.
基金
广东省自然科学基金(2016A030313003)
PDF(1813 KB)
