VNLSTM-PoseNet: A novel deep ConvNet for real-time 6-DOF camera relocalization in urban streets
Image-based relocalization is a renewed interest in outdoor environments, because it is an important problem with many applications. PoseNet introduces Convolutional Neural Network (CNN) for the first time to realize the real-time camera pose solution based on a single image. In order to solve the p...
Main Authors: | , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Taylor & Francis Group
2021-07-01
|
Series: | Geo-spatial Information Science |
Subjects: | |
Online Access: | http://dx.doi.org/10.1080/10095020.2021.1960779 |
id |
doaj-5632c0a145914e78ac49581ade64c193 |
---|---|
record_format |
Article |
spelling |
doaj-5632c0a145914e78ac49581ade64c1932021-10-04T13:57:00ZengTaylor & Francis GroupGeo-spatial Information Science1009-50201993-51532021-07-0124342243710.1080/10095020.2021.19607791960779VNLSTM-PoseNet: A novel deep ConvNet for real-time 6-DOF camera relocalization in urban streetsMing Li0Jiangying Qin1Deren Li2Ruizhi Chen3Xuan Liao4Bingxuan Guo5Wuhan UniversityWuhan UniversityWuhan UniversityWuhan UniversityThe Hong Kong Polytechnic UniversityWuhan UniversityImage-based relocalization is a renewed interest in outdoor environments, because it is an important problem with many applications. PoseNet introduces Convolutional Neural Network (CNN) for the first time to realize the real-time camera pose solution based on a single image. In order to solve the problem of precision and robustness of PoseNet and its improved algorithms in complex environment, this paper proposes and implements a new visual relocation method based on deep convolutional neural networks (VNLSTM-PoseNet). Firstly, this method directly resizes the input image without cropping to increase the receptive field of the training image. Then, the image and the corresponding pose labels are put into the improved Long Short-Term Memory based (LSTM-based) PoseNet network for training and the network is optimized by the Nadam optimizer. Finally, the trained network is used for image localization to obtain the camera pose. Experimental results on outdoor public datasets show our VNLSTM-PoseNet can lead to drastic improvements in relocalization performance compared to existing state-of-the-art CNN-based methods.http://dx.doi.org/10.1080/10095020.2021.1960779camera relocalizationpose regressiondeep convnetrgb imagecamera pose |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Ming Li Jiangying Qin Deren Li Ruizhi Chen Xuan Liao Bingxuan Guo |
spellingShingle |
Ming Li Jiangying Qin Deren Li Ruizhi Chen Xuan Liao Bingxuan Guo VNLSTM-PoseNet: A novel deep ConvNet for real-time 6-DOF camera relocalization in urban streets Geo-spatial Information Science camera relocalization pose regression deep convnet rgb image camera pose |
author_facet |
Ming Li Jiangying Qin Deren Li Ruizhi Chen Xuan Liao Bingxuan Guo |
author_sort |
Ming Li |
title |
VNLSTM-PoseNet: A novel deep ConvNet for real-time 6-DOF camera relocalization in urban streets |
title_short |
VNLSTM-PoseNet: A novel deep ConvNet for real-time 6-DOF camera relocalization in urban streets |
title_full |
VNLSTM-PoseNet: A novel deep ConvNet for real-time 6-DOF camera relocalization in urban streets |
title_fullStr |
VNLSTM-PoseNet: A novel deep ConvNet for real-time 6-DOF camera relocalization in urban streets |
title_full_unstemmed |
VNLSTM-PoseNet: A novel deep ConvNet for real-time 6-DOF camera relocalization in urban streets |
title_sort |
vnlstm-posenet: a novel deep convnet for real-time 6-dof camera relocalization in urban streets |
publisher |
Taylor & Francis Group |
series |
Geo-spatial Information Science |
issn |
1009-5020 1993-5153 |
publishDate |
2021-07-01 |
description |
Image-based relocalization is a renewed interest in outdoor environments, because it is an important problem with many applications. PoseNet introduces Convolutional Neural Network (CNN) for the first time to realize the real-time camera pose solution based on a single image. In order to solve the problem of precision and robustness of PoseNet and its improved algorithms in complex environment, this paper proposes and implements a new visual relocation method based on deep convolutional neural networks (VNLSTM-PoseNet). Firstly, this method directly resizes the input image without cropping to increase the receptive field of the training image. Then, the image and the corresponding pose labels are put into the improved Long Short-Term Memory based (LSTM-based) PoseNet network for training and the network is optimized by the Nadam optimizer. Finally, the trained network is used for image localization to obtain the camera pose. Experimental results on outdoor public datasets show our VNLSTM-PoseNet can lead to drastic improvements in relocalization performance compared to existing state-of-the-art CNN-based methods. |
topic |
camera relocalization pose regression deep convnet rgb image camera pose |
url |
http://dx.doi.org/10.1080/10095020.2021.1960779 |
work_keys_str_mv |
AT mingli vnlstmposenetanoveldeepconvnetforrealtime6dofcamerarelocalizationinurbanstreets AT jiangyingqin vnlstmposenetanoveldeepconvnetforrealtime6dofcamerarelocalizationinurbanstreets AT derenli vnlstmposenetanoveldeepconvnetforrealtime6dofcamerarelocalizationinurbanstreets AT ruizhichen vnlstmposenetanoveldeepconvnetforrealtime6dofcamerarelocalizationinurbanstreets AT xuanliao vnlstmposenetanoveldeepconvnetforrealtime6dofcamerarelocalizationinurbanstreets AT bingxuanguo vnlstmposenetanoveldeepconvnetforrealtime6dofcamerarelocalizationinurbanstreets |
_version_ |
1716844048543645696 |