[1]王泽泓,刘厚泉.基于迁移学习与自适应特征融合的建筑物识别[J].计算机技术与发展,2019,29(12):40-43.[doi:10. 3969 / j. issn. 1673-629X. 2019. 12. 007]
　WANG Ze-hong,LIU Hou-quan.Building Recognition Based on Transfer Learning and Adaptive Feature Fusion[J].,2019,29(12):40-43.[doi:10. 3969 / j. issn. 1673-629X. 2019. 12. 007]
点击复制

基于迁移学习与自适应特征融合的建筑物识别()
分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
29
期数:
2019年12期
页码:
40-43
栏目:
智能、算法、系统工程
出版日期:
2019-12-10

文章信息/Info

Title:
Building Recognition Based on Transfer Learning and Adaptive Feature Fusion
文章编号:
1673-629X(2019)12-0040-04
作者:
王泽泓; 刘厚泉
中国矿业大学 计算机科学与技术学院,江苏 徐州 221116
Author(s):
WANG Ze-hong; LIU Hou-quan
School of Computer Science,China University of Mining and Technology,Xuzhou 221116,China
关键词:
建筑物识别; 卷积神经网络; 迁移学习; 自适应; 特征融合
Keywords:
building recognition; convolutional neural network; transfer learning; adaptive; feature fusion
分类号:
TP301
DOI:
10. 3969 / j. issn. 1673-629X. 2019. 12. 007
摘要:
特定建筑物识别可应用于旅游景点地点的查询。 针对现有特定建筑物识别方法特征提取困难、识别率低等问题,提出一种基于卷积神经网络的特定建筑物识别方法。 针对数据规模小的限制,提出利用预训练的 VGG-16 网络进行迁移学习,以改善网络效果。 为了充分利用 VGG-16 网络中各层提取出的特征,提出自适应特征融合的方法。 该方法针对网络中不同层提取出的特征图的层次不同、尺度不同的特点,给每个特征图设置可学习的权重,进而融合在一起进行预测。通过从网络上爬取的 6 273 张 12 类旅游景点建筑物图片作为数据集,对提出的方法进行验证。 使用 VGG-16 网络训练,最终准确率为 97.86%,处理速度为 296 fps。 利用自适应特征融合改进 VGG-16 网络,改进后最终准确率为 98. 93%,处理速度为 289 fps,比改进前准确率提高 1.07%。
Abstract:
Specific building recognition can be applied to queries about tourist attraction locations. Aiming at the difficulty of feature extraction and low recognition rate of existing specific building recognition methods,a specific building recognition method based on convolutional neural network is proposed. In view of the limitation of small data size,we propose to use the pre-trained VGG-16 network for transfer learning to improve network performance. In order to make full use of the features extracted by each layer in the VGG-16 network,we propose an adaptive feature fusion method in which each feature map is given a learnable weight and integrated to predict according to the characteristics of different layers and scales of feature maps extracted from different layers in the network. The proposed method is verified by taking 6 273 pictures of tourist attractions buildings crawled from the internet as data sets. With VGG-16 network training,the final accuracy rate is 97.86%,and the processing speed is 296 fps. The VGG-16 network is improved by adaptive feature fusion. The improved final accuracy is 98.93% which is improved by 1.07%,and the processing speed is 289 fps.

相似文献/References:

[1]赵洁 李光耀 庞池海 朱恒晔.基于小波变换的卫星遥感地图中建筑物识别[J].计算机技术与发展,2008,(11):243.
　ZHAO Jie,LI Guang-yao,PANG Chi-hai,et al.Research on Building Recognition from Satellite Remote Sensing Image Based on Wavelet[J].,2008,(12):243.
[2]崔凤焦.表情识别算法研究进展与性能比较[J].计算机技术与发展,2018,28(02):145.[doi:10．3969/j．issn．1673－629X．2018．02．031]
　CUI Feng-jiao.Ｒesearch and Performance Comparison of Facial Expression Ｒecognition Algorithm[J].,2018,28(12):145.[doi:10．3969/j．issn．1673－629X．2018．02．031]
[3]张丹丹,李雷. 基于PCANet-RF的人脸检测系统[J].计算机技术与发展,2016,26(02):31.
　ZHANG Dan-dan,LI Lei. Face Detection System Based on PCANet-RF[J].,2016,26(12):31.
[4]陈强锐,谢世朋.基于深度学习的肺部肿瘤检测方法[J].计算机技术与发展,2018,28(04):201.[doi:10.3969/ j. issn.1673-629X.2018.04.043]
　CHEN Qiang-rui,XIE Shi-peng.Lung Cancer Detection Method Based on Deep Learning[J].,2018,28(12):201.[doi:10.3969/ j. issn.1673-629X.2018.04.043]
[5]郭子琰,舒心,刘常燕,等.基于ReLU 函数的卷积神经网络的花卉识别算法[J].计算机技术与发展,2018,28(05):154.[doi:10．3969/j．issn．1673－629X．2018．05．035]
　GUO Ziyan,SHU Xin,LIU Changyan,et al.A Recognition Algorithm of Flower Based on Convolution Neural Network with ReLU Function[J].,2018,28(12):154.[doi:10．3969/j．issn．1673－629X．2018．05．035]
[6]缪宇杰,吴智钧,宫 婧.基于3D 卷积的视频错帧筛选方法[J].计算机技术与发展,2018,28(05):179.[doi:10.3969/ j. issn.1673-629X.2018.05.040]
　MIAO Yu-jie,WU Zhi-jun,GONG Jing.A Wrong Temporal-order Frames Identification Method Based on 3D Convolution[J].,2018,28(12):179.[doi:10.3969/ j. issn.1673-629X.2018.05.040]
[7]吴玉枝,吴志红,熊运余.基于卷积神经网络的小样本车辆检测与识别[J].计算机技术与发展,2018,28(06):1.[doi:10.3969/ j. issn.1673-629X.2018.06.001]
　WU Yu-zhi,WU Zhi-hong,XIONG Yun-yu.Vehicle Detection and Recognition of a Few Samples Based on Convolutional Neural Network[J].,2018,28(12):1.[doi:10.3969/ j. issn.1673-629X.2018.06.001]
[8]李相桥,李晨,田丽华,等.卷积神经网络并行训练的优化研究[J].计算机技术与发展,2018,28(08):12.[doi:10.3969/ j. issn.1673-629X.2018.08.003]
　LI Xiang-qiao,LI Chen,TIAN Li-hua,et al.Research on Optimization of Parallel Training for Convolution Neural Network[J].,2018,28(12):12.[doi:10.3969/ j. issn.1673-629X.2018.08.003]
[9]邓宗平,赵启军,陈虎. 基于深度学习的人脸姿态分类方法[J].计算机技术与发展,2016,26(07):11.
　DEND Zong-ping,ZHAO Qi-jun,CHEN Hu. Face Pose Classification Method Based on Deep Learning[J].,2016,26(12):11.
[10]河海大学 计算机与信息学院,江苏 南京 0098.卷积网络的无监督特征提取对人脸识别的研究[J].计算机技术与发展,2018,28(06):17.[doi:10.3969/ j. issn.1673-629X.2018.06.004]
　DU Bai-sheng.Research on Unsupervised Feature Extraction Based on Convolutional Neural Network for Face Recognition[J].,2018,28(12):17.[doi:10.3969/ j. issn.1673-629X.2018.06.004]

更新日期/Last Update: 2019-12-10