Filters








584 Hits in 3.0 sec

Exploiting uncertainty in regression forests for accurate camera relocalization

Julien Valentin, Matthias Niebner, Jamie Shotton, Andrew Fitzgibbon, Shahram Izadi, Philip Torr
2015 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
by the regression forest; and (ii) leveraging these uncertainties in order to provide for improved relocalization without using explicit models of the scenes.  ...  Recent advances in camera relocalization use predictions from a regression forest to guide the camera pose optimization procedure.  ...  Scene coordinate regression forests for camera relocalization in RGB-D images. In Computer Vision and Pattern Recognition. IEEE, 2013.  ... 
doi:10.1109/cvpr.2015.7299069 dblp:conf/cvpr/ValentinNSFIT15 fatcat:rihvn3n4jjfebaihlccqif7wqq

Exploiting Points and Lines in Regression Forests for RGB-D Camera Relocalization [article]

Lili Meng, Frederick Tung, James J. Little, Julien Valentin, Clarence de Silva
2018 arXiv   pre-print
Line segment features are more robust in these environments. In this work, we propose to jointly exploit points and lines within the framework of uncertainty driven regression forests.  ...  Recent random forests based methods exploit randomly sampled pixel comparison features to predict 3D world locations for 2D image locations to guide the camera pose optimization.  ...  To improve camera relocation accuracy, [4] exploits uncertainty from regression forests by using a Gaussian mixture model in leaf nodes.  ... 
arXiv:1710.10519v3 fatcat:eefiqcqcijh4bcyiipsyp5m66a

Backtracking Regression Forests for Accurate Camera Relocalization [article]

Lili Meng, Jianhui Chen, Frederick Tung, James J. Little, Julien Valentin, Clarence W. de Silva
2017 arXiv   pre-print
Furthermore, we extend the regression forests based methods to use local features in both training and testing stages for outdoor RGB-only applications.  ...  Recent random forests based methods directly predict 3D world locations for 2D image locations to guide the camera pose optimization.  ...  The authors would like to thank The Institute for Computing, Information and Cognitive Systems (ICICS) Member Support Fund in University of British Columbia, Canada.  ... 
arXiv:1710.07965v1 fatcat:z2qj6drjpjbz7lm5fvzxvugm2m

VidLoc: A Deep Spatio-Temporal Model for 6-DoF Video-Clip Relocalization [article]

Ronald Clark, Sen Wang, Andrew Markham, Niki Trigoni, Hongkai Wen
2017 arXiv   pre-print
Machine learning techniques, namely convolutional neural networks (CNN) and regression forests, have recently shown great promise in performing 6-DoF localization of monocular images.  ...  In this paper we propose a recurrent model for performing 6-DoF localization of video-clips.  ...  However, for global camera relocalization as we are concerned with in this paper, this assumption is unlikely.  ... 
arXiv:1702.06521v2 fatcat:7dj5wwhhvfaati4k7i5ynowb4q

VidLoc: A Deep Spatio-Temporal Model for 6-DoF Video-Clip Relocalization

Ronald Clark, Sen Wang, Andrew Markham, Niki Trigoni, Hongkai Wen
2017 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
Machine learning techniques, namely convolutional neural networks (CNN) and regression forests, have recently shown great promise in performing 6-DoF localization of monocular images.  ...  In this paper we propose a recurrent model for performing 6-DoF localization of video-clips.  ...  However, for global camera relocalization as we are concerned with in this paper, this assumption is unlikely.  ... 
doi:10.1109/cvpr.2017.284 dblp:conf/cvpr/ClarkWMTW17 fatcat:ltmhuc3vrvg7rbialdlmb3uace

DynaLoc: Real-Time Camera Relocalization from a Single RGB Image in Dynamic Scenes based on an Adaptive Regression Forest

Nam-Duong Duong, Amine Kacete, Catherine Soladie, Pierre-Yves Richard, Jérôme Royan
2020 Proceedings of the 15th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications  
In this paper, we propose an adaptive regression forest and apply it to our DynaLoc, a real-time camera relocalization approach from a single RGB image in dynamic environments.  ...  Our adaptive regression forest is able to fine tune and update continuously itself from evolving data in real-time.  ...  We apply our adaptive regression forest to our DynaLoc, a realtime and accurate camera relocalization from a singe RGB image in dynamic scenes with moving objects.  ... 
doi:10.5220/0008902906430653 dblp:conf/visapp/DuongKSRR20 fatcat:2ivpqcsbnfckhj7rzb7btdb3hm

Real-time Direct Monocular SLAM with Learning-based Confidence Estimation

Weiqi Zhang, Zifei Yan, Gang Xiao, Aidi Feng, Wangmeng Zuo
2019 IEEE Access  
In GCP-SLAM, a random regression forest is trained off-line with pre-defined confidence measures for learning confidence and detecting the ground control points (GCPs).  ...  However, feature-based methods usually have more accurate camera localization results than most direct methods, though direct methods can work better in a textureless environment.  ...  In camera localization, the selected GCPs are exploited for minimizing the photometric error.  ... 
doi:10.1109/access.2019.2928578 fatcat:f6zkvtieyfh6xgwbiq3dcaijla

ARtwin: Map pivot format specification

Nam Duong Duong, Jérôme Royan, Michal Polic, Nischita Sudharsan
2020 Zenodo  
This project has received funding from the European Union's Horizon 2020 Framework Programme for research and Innovation under Grant Agreement no 856994.  ...  Camera pose regression is an end-to-end approach for camera relocalization. It regresses directly camera pose from each input data.  ...  [118] maps parameters between regression forests and neural networks to leverage the performance benefits of neural networks for dense regression while retaining the efficiency of random forests for  ... 
doi:10.5281/zenodo.4381408 fatcat:pn2ioshymfcrrkbuhxg72kr7zy

Scene Coordinate Regression with Angle-Based Reprojection Loss for Camera Relocalization [article]

Xiaotian Li, Juha Ylioinas, Jakob Verbeek, Juho Kannala
2018 arXiv   pre-print
Image-based camera relocalization is an important problem in computer vision and robotics.  ...  Recent works utilize convolutional neural networks (CNNs) to regress for pixels in a query image their corresponding 3D world coordinates in the scene.  ...  This work has also been partially supported by the grant Deep in France (ANR16-CE23-0006) and LabEx PERSYVAL (ANR-11-LABX0025-01).  ... 
arXiv:1808.04999v2 fatcat:462mrjhmyzeo5lz5jwuuulxatm

Scene Coordinate Regression with Angle-Based Reprojection Loss for Camera Relocalization [chapter]

Xiaotian Li, Juha Ylioinas, Jakob Verbeek, Juho Kannala
2019 Lecture Notes in Computer Science  
Image-based camera relocalization is an important problem in computer vision and robotics.  ...  Recent works utilize convolutional neural networks (CNNs) to regress for pixels in a query image their corresponding 3D world coordinates in the scene.  ...  For example, in [17] , the authors explore Bayesian Neural Networks to produce relocalization uncertainty of the predicted pose.  ... 
doi:10.1007/978-3-030-11015-4_19 fatcat:ntesaqmj6rhifdxj3au5ozv7qe

Camera Relocalization by Exploiting Multi-View Constraints for Scene Coordinates Regression

Ming Cai, Huangying Zhan, Chamara Saroj Weerasekera, Kejie Li, Ian Reid
2019 2019 IEEE/CVF International Conference on Computer Vision Workshop (ICCVW)  
We propose a method for learning a scene coordinate regression model to perform accurate camera relocalization in a known environment from a single RGB image.  ...  For the warp error we explore both RGB values, and deep learned features, as the basis for the error.  ...  [25] exploits the uncertainty in the estimate from the Random Forest to benefit the pose optimization.  ... 
doi:10.1109/iccvw.2019.00469 dblp:conf/iccvw/CaiZWLR19 fatcat:dvtj5vmfuvgjzfaaeg4dzltjpe

KFNet: Learning Temporal Camera Relocalization using Kalman Filtering [article]

Lei Zhou, Zixin Luo, Tianwei Shen, Jiahui Zhang, Mingmin Zhen, Yao Yao, Tian Fang, Long Quan
2020 arXiv   pre-print
In this work, we improve the temporal relocalization method by using a network architecture that incorporates Kalman filtering (KFNet) for online camera relocalization.  ...  In particular, KFNet extends the scene coordinate regression problem to the time domain in order to recursively establish 2D and 3D correspondences for the pose determination.  ...  Bayesian Formulation This section presents the Bayesian formulation of recursive scene coordinate regression in the time domain for temporal camera relocalization.  ... 
arXiv:2003.10629v1 fatcat:yohcxo4wanfzrdvtg7rt52fe7u

Decoupling Features and Coordinates for Few-shot RGB Relocalization [article]

Siyan Dong, Songyin Wu, Yixin Zhuang, Kai Xu, Shanghang Zhang, Baoquan Chen
2020 arXiv   pre-print
Cross-scene model adaption is crucial for camera relocalization in real scenarios.  ...  To address this issue, we approach camera relocalization with a decoupled solution where feature extraction, coordinate regression, and pose estimation are performed separately.  ...  For those two-step camera relocalization methods, pixelwise feature extraction and coordinate regression are performed in an integrated fashion.  ... 
arXiv:1911.11534v3 fatcat:lzh4cnu4ora57o53kv43ssoke4

Image-Based Localization Using Hourglass Networks

Iaroslav Melekhov, Juha Ylioinas, Juho Kannala, Esa Rahtu
2017 2017 IEEE International Conference on Computer Vision Workshops (ICCVW)  
In this paper, we propose an encoder-decoder convolutional neural network (CNN) architecture for estimating camera pose (orientation and location) from a single RGBimage.  ...  The architecture has a hourglass shape consisting of a chain of convolution and up-convolution layers followed by a regression part.  ...  [29] further improved the method by exploiting uncertainty in the model in order to move from sole point estimates to predict also their uncertainties for more robust continuous pose optimization.  ... 
doi:10.1109/iccvw.2017.107 dblp:conf/iccvw/MelekhovYKR17 fatcat:2xh4n4dckjhazgy7kc7hdfx3fy

Image-based Localization using Hourglass Networks [article]

Iaroslav Melekhov, Juha Ylioinas, Juho Kannala, Esa Rahtu
2017 arXiv   pre-print
In this paper, we propose an encoder-decoder convolutional neural network (CNN) architecture for estimating camera pose (orientation and location) from a single RGB-image.  ...  The architecture has a hourglass shape consisting of a chain of convolution and up-convolution layers followed by a regression part.  ...  [30] further improved the method by exploiting uncertainty in the model in order to move from sole point estimates to predict also their uncertainties for more robust continuous pose optimization.  ... 
arXiv:1703.07971v3 fatcat:csrnilyconawjlj7rkhgogjlfu
« Previous Showing results 1 — 15 out of 584 results