Global Optimal Structured Embedding Learning for Remote Sensing Image Retrieval
A rich line of works focus on designing elegant loss functions under the deep metric learning (DML) paradigm to learn a discriminative embedding space for remote sensing image retrieval (RSIR). Essentially, such embedding space could efficiently distinguish deep feature descriptors. So far, most exi...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2020-01-01
|
Series: | Sensors |
Subjects: | |
Online Access: | https://www.mdpi.com/1424-8220/20/1/291 |
id |
doaj-7028afdab3f74d24a4a11f14b8744e6a |
---|---|
record_format |
Article |
spelling |
doaj-7028afdab3f74d24a4a11f14b8744e6a2020-11-25T02:20:44ZengMDPI AGSensors1424-82202020-01-0120129110.3390/s20010291s20010291Global Optimal Structured Embedding Learning for Remote Sensing Image RetrievalPingping Liu0Guixia Gou1Xue Shan2Dan Tao3Qiuzhan Zhou4College of Computer Science and Technology, Jilin University, Changchun 130012, ChinaCollege of Computer Science and Technology, Jilin University, Changchun 130012, ChinaCollege of Computer Science and Technology, Jilin University, Changchun 130012, ChinaSchool of Electronic and Information Engineering, Beijing Jiaotong University, Beijing 100044, ChinaCollege of Communication Engineering, Jilin University, Changchun 130012, ChinaA rich line of works focus on designing elegant loss functions under the deep metric learning (DML) paradigm to learn a discriminative embedding space for remote sensing image retrieval (RSIR). Essentially, such embedding space could efficiently distinguish deep feature descriptors. So far, most existing losses used in RSIR are based on triplets, which have disadvantages of local optimization, slow convergence and insufficient use of similarity structure in a mini-batch. In this paper, we present a novel DML method named as global optimal structured loss to deal with the limitation of triplet loss. To be specific, we use a softmax function rather than a hinge function in our novel loss to realize global optimization. In addition, we present a novel optimal structured loss, which globally learn an efficient deep embedding space with mined informative sample pairs to force the positive pairs within a limitation and push the negative ones far away from a given boundary. We have conducted extensive experiments on four public remote sensing datasets and the results show that the proposed global optimal structured loss with pairs mining scheme achieves the state-of-the-art performance compared with the baselines.https://www.mdpi.com/1424-8220/20/1/291remote sensing image retrievalconvolutional neural networkdeep metric learningglobal optimization |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Pingping Liu Guixia Gou Xue Shan Dan Tao Qiuzhan Zhou |
spellingShingle |
Pingping Liu Guixia Gou Xue Shan Dan Tao Qiuzhan Zhou Global Optimal Structured Embedding Learning for Remote Sensing Image Retrieval Sensors remote sensing image retrieval convolutional neural network deep metric learning global optimization |
author_facet |
Pingping Liu Guixia Gou Xue Shan Dan Tao Qiuzhan Zhou |
author_sort |
Pingping Liu |
title |
Global Optimal Structured Embedding Learning for Remote Sensing Image Retrieval |
title_short |
Global Optimal Structured Embedding Learning for Remote Sensing Image Retrieval |
title_full |
Global Optimal Structured Embedding Learning for Remote Sensing Image Retrieval |
title_fullStr |
Global Optimal Structured Embedding Learning for Remote Sensing Image Retrieval |
title_full_unstemmed |
Global Optimal Structured Embedding Learning for Remote Sensing Image Retrieval |
title_sort |
global optimal structured embedding learning for remote sensing image retrieval |
publisher |
MDPI AG |
series |
Sensors |
issn |
1424-8220 |
publishDate |
2020-01-01 |
description |
A rich line of works focus on designing elegant loss functions under the deep metric learning (DML) paradigm to learn a discriminative embedding space for remote sensing image retrieval (RSIR). Essentially, such embedding space could efficiently distinguish deep feature descriptors. So far, most existing losses used in RSIR are based on triplets, which have disadvantages of local optimization, slow convergence and insufficient use of similarity structure in a mini-batch. In this paper, we present a novel DML method named as global optimal structured loss to deal with the limitation of triplet loss. To be specific, we use a softmax function rather than a hinge function in our novel loss to realize global optimization. In addition, we present a novel optimal structured loss, which globally learn an efficient deep embedding space with mined informative sample pairs to force the positive pairs within a limitation and push the negative ones far away from a given boundary. We have conducted extensive experiments on four public remote sensing datasets and the results show that the proposed global optimal structured loss with pairs mining scheme achieves the state-of-the-art performance compared with the baselines. |
topic |
remote sensing image retrieval convolutional neural network deep metric learning global optimization |
url |
https://www.mdpi.com/1424-8220/20/1/291 |
work_keys_str_mv |
AT pingpingliu globaloptimalstructuredembeddinglearningforremotesensingimageretrieval AT guixiagou globaloptimalstructuredembeddinglearningforremotesensingimageretrieval AT xueshan globaloptimalstructuredembeddinglearningforremotesensingimageretrieval AT dantao globaloptimalstructuredembeddinglearningforremotesensingimageretrieval AT qiuzhanzhou globaloptimalstructuredembeddinglearningforremotesensingimageretrieval |
_version_ |
1724870242634563584 |