[1]朱周华,高 凡.基于深度学习的局部实例搜索[J].计算机技术与发展,2020,30(09):36-42.[doi:10. 3969 / j. issn. 1673-629X. 2020. 09. 007]
　ZHU Zhou-hua,GAO Fan.Local Instance Search Based on Deep Learning[J].,2020,30(09):36-42.[doi:10. 3969 / j. issn. 1673-629X. 2020. 09. 007]
点击复制

基于深度学习的局部实例搜索()
分享到：

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

卷:
30
期数:
2020年09期
页码:
36-42
栏目:
智能、算法、系统工程
出版日期:
2020-09-10

文章信息/Info

Title:
Local Instance Search Based on Deep Learning
文章编号:
1673-629X(2020)09-0036-07
作者:
朱周华; 高 凡
西安科技大学 通信与信息工程学院,陕西 西安 710054
Author(s):
ZHU Zhou-hua; GAO Fan
School of Communication and Information Engineering,Xi’an University of Science and Technology,Xi’an 710054,China
关键词:
深度学习; 局部实例搜索; 区域特征; 微调; 特征匹配
Keywords:
deep learning; local instance search; regional features; fine-tuning; feature matching
分类号:
TP391
DOI:
10. 3969 / j. issn. 1673-629X. 2020. 09. 007
摘要:
针对传统实例搜索方法准确率和视觉相似度低下的问题,提出一种利用卷积神经网络提取图像全局特征和区域特征的实例搜索方法。 该方法经过初步搜索、重排和查询扩展三个阶段实现实例搜索任务,通过微调策略和在重排阶段对特征匹配方法的改进进一步提高检索性能,并将其应用到局部实例搜索任务,即利用残缺图像检索得到整幅图像,在此基础之上,加入在线检索功能。 在 Oxford 5k 和 Paris 6k 两个公开数据集上进行实验验证,结果表明,整幅图像的检索 mAP值和视觉相似度都得到了很大提升,局部实例检索的 mAP 值均高于其他文献中整幅图像的检索,仅比文中整幅图像的检索低 0.032。 因此,提出的实例搜索方法不仅提高了实例搜索的准确率,也增强了目标定位的准确性,同时很好地解决了局部实例搜索问题。
Abstract:
In order to solve the problem of low accuracy and visual similarity of traditional instance search methods,an instance search method for extracting global and regional features of images using convolutional neural networks is proposed. The method realizes instance search task through three stages:filtering stage,spatial re-ranking and query expansion. The retrieval performance is further improved by fine-tuning strategy and the improvement of the feature matching method in the re-ranking stage. It is applied to the local instance search task,that is,using the incomplete image retrieval to obtain the whole image. On this basis,the online search function is added. Experiments are carried out on two public datasets of Oxford building 5k and Paris building 6k that the mAP value and visual similarity of the whole image are greatly improved. The mAP value of the local instance retrieval is higher than that of other literatures. The retrieval of images is only 0.032 lower than the retrieval of the entire image. Therefore,the proposed method not only improves the accuracy of instance search,but also enhances the accuracy of target location,and solves the problem of local instance search.

相似文献/References:

[1]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(09):201.[doi:10.3969/ j. issn.1673-629X.2018.04.043]
[2]施泽浩,赵启军.基于全卷积网络的目标检测算法[J].计算机技术与发展,2018,28(05):55.[doi:10.3969/j.issn.1673－629X.2018.05.013]
　SHI Ze-hao,ZHAO Qi-jun.Object Detection Algorithm Based on Fully Convolutional Neural Network[J].,2018,28(09):55.[doi:10.3969/j.issn.1673－629X.2018.05.013]
[3]黄法秀,张世杰,吴志红,等.数据增广下的人脸识别研究[J].计算机技术与发展,2020,30(03):67.[doi:10. 3969 / j. issn. 1673-629X. 2020. 03. 013]
　HUANG Fa-xiu,ZHANG Shi-jie,WU Zhi-hong,et al.Research on Face Recognition Based on Data Augmentation[J].,2020,30(09):67.[doi:10. 3969 / j. issn. 1673-629X. 2020. 03. 013]
[4]陈浩翔,蔡建明,刘铿然,等. 手写数字深度特征学习与识别[J].计算机技术与发展,2016,26(07):19.
　CHEN Hao-xiang,CAI Jian-ming,LIU Keng-ran,et al. Deep Learning and Recognition of Handwritten Numeral Features[J].,2016,26(09):19.
[5]高翔,陈志,岳文静,等.基于视频场景深度学习的人物语义识别模型[J].计算机技术与发展,2018,28(06):53.[doi:10.3969/ j. issn.1673-629X.2018.06.012]
　GAO Xiang,CHEN Zhi,YUE Wen-jing,et al.Human Semantic Recognition Model Based on Video Scene Deep Learning[J].,2018,28(09):53.[doi:10.3969/ j. issn.1673-629X.2018.06.012]
[6]贺飞翔,赵启军. 基于深度学习的头部姿态估计[J].计算机技术与发展,2016,26(11):1.
　HE Fei-xiang,ZHAO Qi-jun. Head Pose Estimation Based on Deep Learning[J].,2016,26(09):1.
[7]徐 融,邱晓晖.一种改进的 YOLO V3 目标检测方法[J].计算机技术与发展,2020,30(07):30.[doi:10. 3969 / j. issn. 1673-629X. 2020. 07. 007]
　XU Rong,QIU Xiao-hui.An Improved YOLO V3 Object Detection[J].,2020,30(09):30.[doi:10. 3969 / j. issn. 1673-629X. 2020. 07. 007]
[8]曾志平[] [],萧海东[],张新鹏[]. 基于DBN的金融时序数据建模与决策[J].计算机技术与发展,2017,27(04):1.
　ZENG Zhi-ping[] [],XIAO Hai-dong[],ZHANG Xin-peng[]. Modeling and Decision-making of Financial Time Series Data with DBN[J].,2017,27(09):1.
[9]李全兵,文 钊*,田艳梅*,等.基于 WGAN 的音频关键词识别研究[J].计算机技术与发展,2021,31(08):26.[doi:10. 3969 / j. issn. 1673-629X. 2021. 08. 005]
　LI Quan-bing,WEN Zhao *,TIAN Yan-mei *,et al.Research on Audio Keywords Recognition Based on WassersteinGenerative Adversarial Network[J].,2021,31(09):26.[doi:10. 3969 / j. issn. 1673-629X. 2021. 08. 005]
[10]李宏林. 分析式纹理合成技术及其在深度学习的应用[J].计算机技术与发展,2017,27(11):7.
　LI Hong-lin. Analyzed Texture-synthesis Techniques and Their Applications in Deep Learning[J].,2017,27(09):7.

更新日期/Last Update: 2020-09-10