Filters








622 Hits in 8.8 sec

Supervising the New with the Old: Learning SFM from SFM [chapter]

Maria Klodt, Andrea Vedaldi
2018 Lecture Notes in Computer Science  
Recent work has demonstrated that it is possible to learn deep neural networks for monocular depth and ego-motion estimation from unlabelled video sequences, an interesting theoretical development with  ...  We also propose to build on dozens of years of experience in developing handcrafted structure-from-motion (SFM) algorithms.  ...  We compare results from SfMLearner as stated in the paper and on the website, to the proposed method with uncertainties and depth and pose supervision from SfM.  ... 
doi:10.1007/978-3-030-01249-6_43 fatcat:h4cpjsuucbg5bnukaugk4akyzi

Fusing the Old with the New: Learning Relative Camera Pose with Geometry-Guided Uncertainty [article]

Bingbing Zhuang, Manmohan Chandraker
2021 arXiv   pre-print
The question of how to integrate predictions from deep neural networks (DNNs) and solutions from geometric solvers, such as the 5-point algorithm, has as yet remained under-explored.  ...  Learning methods for relative camera pose estimation have been developed largely in isolation from classical geometric approaches.  ...  In the context of SFM, Klodt and Vedaldi [27] utilize uncertainty learning to handle the varying reliability of geometric SFM when applied as reference ground truth to supervise the SFM learning.  ... 
arXiv:2104.08278v1 fatcat:cadwd6ghivasdgl6nxv3nklkga

Towards Continual, Online, Self-Supervised Depth [article]

Muhammad Umar Karim Khan
2022 arXiv   pre-print
Online adaptation, where the neural network trains while deployed, with self-supervised learning provides a convenient solution as the network can learn from the scene where it is deployed without external  ...  The proposed approach is more inline with the artificial general intelligence paradigm as the neural network learns continually with no supervision.  ...  Knowledge transfer in similar tasks is dealt with in [34] . [35] deals with continual learning where new tasks may contain samples from old tasks. [36] gates layers based on the current task and also  ... 
arXiv:2103.00369v3 fatcat:jhzpki73zvczdp5z2npw3xsbtu

Technical Challenges in Developing Software to Collect Twitter Data

Daniel Chudnov, Daniel Kerchner, Ankushi Sharma, Laura Wrubel
2014 Code4Lib Journal  
Expanding the project from a research prototype to a more widely useful application has presented a number of technical challenges, including changes in the Twitter API, supervision of simultaneous streaming  ...  Over the past two years, George Washington University Libraries developed Social Feed Manager (SFM), a Python and Django-based application for collecting social media data from Twitter.  ...  Among many Python libraries evaluated, Tweepy had excellent coverage of the old API and an active development branch for the new API well before Twitter disabled the old API.  ... 
doaj:5fa5d3aecc4542f5b9d3a3ae1b5f0a42 fatcat:4ygwu25sv5ezxkkaqvp46zit4m

From Coarse to Fine: Robust Hierarchical Localization at Large Scale

Paul-Edouard Sarlin, Cesar Cadena, Roland Siegwart, Marcin Dymczyk
2019 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
By leveraging learned descriptors, our method achieves remarkable localization robustness across large variations of appearance and sets a new state-of-the-art on two challenging benchmarks for large-scale  ...  State-of-the-art methods not only struggle with such scenarios, but are often too resource intensive for certain real-time applications.  ...  We thank the reviewers for their valuable comments, Torsten Sattler for helping to evaluate the localization, and Eduard Trulls for providing support for the SfM dataset.  ... 
doi:10.1109/cvpr.2019.01300 dblp:conf/cvpr/SarlinCSD19 fatcat:kiaadibacne7tmiyfgsmtqhnt4

From Coarse to Fine: Robust Hierarchical Localization at Large Scale [article]

Paul-Edouard Sarlin, Cesar Cadena, Roland Siegwart, Marcin Dymczyk
2019 arXiv   pre-print
By leveraging learned descriptors, our method achieves remarkable localization robustness across large variations of appearance and sets a new state-of-the-art on two challenging benchmarks for large-scale  ...  State-of-the-art methods not only struggle with such scenarios, but are often too resource intensive for certain real-time applications.  ...  We thank the reviewers for their valuable comments, Torsten Sattler for helping to evaluate the localization, and Eduard Trulls for providing support for the SfM dataset.  ... 
arXiv:1812.03506v2 fatcat:7ccgbre2ezcn3oj2526jxbuvfy

On the Limits of Pseudo Ground Truth in Visual Camera Re-localisation [article]

Eric Brachmann, Martin Humenberger, Carsten Rother, Torsten Sattler
2021 arXiv   pre-print
Popular choices include Structure-from-Motion (SfM) and Simultaneous-Localisation-and-Mapping (SLAM) using additional sensors like depth cameras if available.  ...  This paper analyzes two widely used re-localisation datasets and shows that evaluation outcomes indeed vary with the choice of the reference algorithm.  ...  We compare the SfM pGT with point clouds obtained by triangulating the scenes from the original (RGB-)D pGT.  ... 
arXiv:2109.00524v1 fatcat:spz6mouvmfd5bjkztimbmiay6i

Adversarial Inverse Graphics Networks: Learning 2D-to-3D Lifting and Image-to-Image Translation from Unpaired Supervision [article]

Hsiao-Yu Fish Tung, Adam W. Harley, William Seto, Katerina Fragkiadaki
2017 arXiv   pre-print
Learning such mappings from unlabelled data, or improving upon supervised models by exploiting unlabelled data, remains elusive.  ...  We propose Adversarial Inverse Graphics networks (AIGNs): weakly supervised neural network models that combine feedback from rendering their predictions, with distribution matching between their predictions  ...  Self-supervised learning of SfM.  ... 
arXiv:1705.11166v3 fatcat:xe4mr3ajqbd35c23xyiyxac3ym

Are Large-Scale 3D Models Really Necessary for Accurate Visual Localization?

Torsten Sattler, Akihiko Torii, Josef Sivic, Marc Pollefeys, Hajime Taira, Masatoshi Okutomi, Tomas Pajdla
2017 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
Using these poses, we show that combining image-based methods with local reconstructions results in a higher pose accuracy compared to state-of-the-art structure-based methods, albeight at higher run-time  ...  They are often considered inaccurate since they only approximate the positions of the cameras.  ...  Structure-based localization approaches naturally benefit from deep learning by replacing handcrafted feature detectors and descriptors such as SIFT [56] with learned alternatives [45] , [60] , [61  ... 
doi:10.1109/cvpr.2017.654 dblp:conf/cvpr/SattlerTSPTOP17 fatcat:zdovhsf5g5c67dlky5lneb5zsm

Fluvial and aquatic applications of Structure from Motion photogrammetry and unmanned aerial vehicle/drone technology

Jonathan L. Carrivick, Mark W. Smith
2018 WIREs Water  
0000-0002-9286-5348 and Smith, MW orcid.org/0000-0003-4361-9527 (2019) Fluvial and aquatic applications of Structure from Motion photogrammetry and unmanned aerial vehicle/drone technology.  ...  Acknowledgments Very helpful reviews from Patrice Carbonneau and Mike James improved this overview article.  ...  There is great potential for each to learn from the ging solutions.  ... 
doi:10.1002/wat2.1328 fatcat:mxztbt7mgvdlbipu4klbajkgsu

Impact of prevalence and case distribution in lab-based diagnostic imaging studies

Brandon D. Gallas, Weijie Chen, Elodia Cole, Robert Ochs, Nicholas Petrick, Etta D. Pisano, Berkman Sahiner, Frank W. Samuelson, Kyle J. Myers
2019 Journal of Medical Imaging  
Our retrospective reader studies compared full-field digital mammography (FFDM) to screen-film mammography (SFM) for women with dense breasts.  ...  Mammograms were acquired from the prospective Digital Mammographic Imaging Screening Trial.  ...  on the study design, analyses, and results.  ... 
doi:10.1117/1.jmi.6.1.015501 pmid:30713851 pmcid:PMC6340399 fatcat:xymcaf3fgnhqfbwxue4e72atni

Co-Attention for Conditioned Image Matching [article]

Olivia Wiles, Sebastien Ehrhardt, Andrew Zisserman
2021 arXiv   pre-print
We propose a new approach to determine correspondences between image pairs in the wild under large changes in illumination, viewpoint, context, and material.  ...  To achieve this, we introduce (i) a spatial attention mechanism (a co-attention module, CoAM) for conditioning the learned features on both images, and (ii) a distinctiveness score used to choose the best  ...  Structure from Motion (SfM) The objective here is to evaluate the correspondences obtained with our model for the task of 3D reconstruction. SfM Dataset.  ... 
arXiv:2007.08480v2 fatcat:7fpd7b5rrje4lck2iodmlgxfvi

Comparative Analysis of Digital Models of Objects of Cultural Heritage Obtained by the "3D SLS" and "SfM" Methods

Marcin Barszcz, Jerzy Montusiewicz, Magdalena Paśnikowska-Łukaszuk, Anna Sałamacha
2021 Applied Sciences  
The obtained results show that the quality of the 3D models generated with SfM differs from the models from the technology (3D SLS), but they may be placed in the galleries of the vitrual museum.  ...  The obtained models from SfM do not have information about their size, which means that they are not fully suitable for archiving purposes of cultural heritage, unlike the models from SLS.  ...  new Director of the Afrasiyab Museum, as well as the employees of the Museum for providing a collection of photos of objects from the exhibition.  ... 
doi:10.3390/app11125321 fatcat:jk4ot2yplbaobf5vavhq526dnq

UV3D: Underwater Video Stream 3D Reconstruction Based on Efficient Global SFM

Yanli Chen, Qiushi Li, Shenghua Gong, Jun Liu, Wenxue Guan
2022 Applied Sciences  
Firstly, we propose an adaptive threshold key frame extraction algorithm based on clustering, which extracts key frames from the video stream as structure from motion (SFM) image sequences.  ...  Additionally, we choose global SFM to construct the scene and propose a faster rotation averaging method, the least trimmed square rotation averaging (LTS-RA) method, based on the least trimmed squares  ...  Conflicts of Interest: The authors declare no conflict of interest.  ... 
doi:10.3390/app12125918 fatcat:eoikwgl5nnag5dnlwxbwpv5gwm

Modelling the Surface of Racing Vessel's Hull by Laser Scanning and Digital Photogrammetry

Karol Bartoš, Katarína Pukanská, Peter Repáň, Ľubomír Kseňak, Janka Sabová
2019 Remote Sensing  
Structure-from-Motion (SfM) method.  ...  Nevertheless, although labour intensive, the SfM photogrammetry can be considered the most suitable method if the correct procedures are followed.  ...  of e-learning approaches in education."  ... 
doi:10.3390/rs11131526 fatcat:4pqbg3e7tngynabadlbflcpsoi
« Previous Showing results 1 — 15 out of 622 results