Filters








659 Hits in 3.9 sec

Visual Estimation of Building Condition with Patch-level ConvNets

David Koch, Miroslav Despotovic, Muntaha Sakeena, Mario Döller, Matthias Zeppelzauer
2018 Proceedings of the 2018 ACM Workshop on Multimedia for Real Estate Tech - RETech'18  
To this end, we develop a multi-scale patch-based pattern extraction approach and combine it with convolutional neural networks to estimate building condition from visual clues.  ...  Our evaluation shows that visually estimated building condition can serve as a proxy for condition estimates by appraisers.  ...  Experiments with a large dataset show that useful visual clues for the estimation of condition can be extracted automatically and that the estimated condition has a positive impact on price estimation.  ... 
doi:10.1145/3210499.3210526 dblp:conf/mir/KochDSDZ18 fatcat:7haoeulyqbhm5avqom6db5ewly

Automatic Feature Learning for Robust Shadow Detection

Salman Hameed Khan, Mohammed Bennamoun, Ferdous Sohel, Roberto Togneri
2014 2014 IEEE Conference on Computer Vision and Pattern Recognition  
The 7-layer network architecture of each ConvNet consists of alternating convolution and sub-sampling layers.  ...  Our proposed framework consistently performed better than the state-of-the-art on all major shadow databases collected under a variety of conditions.  ...  Feature Learning with ConvNets We employ multiple ConvNets for feature learning along the boundaries and at the super-pixel level.  ... 
doi:10.1109/cvpr.2014.249 dblp:conf/cvpr/KhanBST14 fatcat:fwkhclwvnfc7tmuy7ygkfckvqm

Learning Generative Models with Visual Attention [article]

Yichuan Tang, Nitish Srivastava, Ruslan Salakhutdinov
2015 arXiv   pre-print
Inspired by the visual attention models in computational neuroscience and the need of object-centric data for generative models, we describe for generative learning framework using attentional mechanisms  ...  A ConvNet is employed to provide good initializations during posterior inference which is based on Hamiltonian Monte Carlo.  ...  CMU Multi-PIE training set in order to pretrain our ConvNet.  ... 
arXiv:1312.6110v3 fatcat:mfjbo6gwerdbphufxeidycswia

3DMatch: Learning Local Geometric Descriptors from RGB-D Reconstructions [article]

Andy Zeng, Shuran Song, Matthias Nießner, Matthew Fisher, Jianxiong Xiao, Thomas Funkhouser
2017 arXiv   pre-print
In this paper, we present 3DMatch, a data-driven model that learns a local volumetric patch descriptor for establishing correspondences between partial 3D data.  ...  These difficulties limit the performance of current state-of-art methods, which are typically based on histograms over geometric properties.  ...  Matthias Nießner is a member of the Max Planck Center for Visual Computing and Communications (MPC-VCC). We gratefully acknowledge the support of NVIDIA and Intel for hardware donations.  ... 
arXiv:1603.08182v3 fatcat:kmxnnqwd5zgkvj3rlu3rh7iilm

From ImageNet to Mining: Adapting Visual Object Detection with Minimal Supervision [chapter]

Alex Bewley, Ben Upcroft
2016 Springer Tracts in Advanced Robotics  
This paper presents visual detection and classification of light vehicles and personnel on a mine site.  ...  We exploit the abundance of background-only images to train a k-means classifier to complement the ConvNet.  ...  Acknowledgement also goes to the high performance computing group at Queensland University of Technology for both support and use of their services when conducting the experiments in this paper.  ... 
doi:10.1007/978-3-319-27702-8_33 fatcat:yg7hjxs5kndqjdugj5tkyrufty

Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks [article]

Emily Denton, Soumith Chintala, Arthur Szlam, Rob Fergus
2015 arXiv   pre-print
At each level of the pyramid, a separate generative convnet model is trained using the Generative Adversarial Nets (GAN) approach (Goodfellow et al.).  ...  In this paper we introduce a generative parametric model capable of producing high quality samples of natural images.  ...  Against this background, our proposed approach makes a significant advance in that it is straightforward to train and sample from, with the resulting samples showing a surprising level of visual fidelity  ... 
arXiv:1506.05751v1 fatcat:jmlxqhrtmvccrfl2gmdtyd2pou

PoseNet: A Convolutional Network for Real-Time 6-DOF Camera Relocalization [article]

Alex Kendall, Matthew Grimes, Roberto Cipolla
2016 arXiv   pre-print
This is achieved using an efficient 23 layer deep convnet, demonstrating that convnets can be used to solve complicated out of image plane regression problems.  ...  We show the convnet localizes from high level features and is robust to difficult lighting, motion blur and different camera intrinsics where point based SIFT registration fails.  ...  (a) Relocalization with increasing levels of motion blur. The system is able to recognize the pose as high level features such as the contour outline still exist.  ... 
arXiv:1505.07427v4 fatcat:z5q36q3aezamhp2tis3o5lty24

Convolutional neural networks for high throughput screening of catalyst layer inks for polymer electrolyte fuel cells

Mohammad J. Eslamibidgoli, Fabian P. Tipp, Jenia Jitsev, Jasna Jankovic, Michael H. Eikerling, Kourosh Malek
2021 RSC Advances 11(51)  
The structural building blocks of catalyst layers are formed during the processing and application of catalyst inks.  ...  This article presents the first application of ConvNets for the high throughput screening of transmission electron microscopy images at the ink stage.  ...  (http://www.gauss-centre.eu) for funding this project by providing computing time on the GCS Supercomputer JUWELS 66 at Jülich Supercomputing Centre (JSC) of Forschunszentrum Jülich.  ... 
doi:10.18154/rwth-2021-12097 fatcat:nvfcjgavqvb23eqks35kjncsgq

3DMatch: Learning Local Geometric Descriptors from RGB-D Reconstructions

Andy Zeng, Shuran Song, Matthias NieBner, Matthew Fisher, Jianxiong Xiao, Thomas Funkhouser
2017 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
In this paper, we present 3DMatch, a data-driven model that learns a local volumetric patch descriptor for establishing correspondences between partial 3D data.  ...  These difficulties limit the performance of current state-of-art methods, which are typically based on histograms over geometric properties.  ...  Matthias Nießner is a member of the Max Planck Center for Visual Computing and Communications (MPC-VCC). We gratefully acknowledge the support of NVIDIA and Intel for hardware donations.  ... 
doi:10.1109/cvpr.2017.29 dblp:conf/cvpr/ZengSNFXF17 fatcat:7pbrzfzp65aszm3owiayxnnrqm

SimNets: A Generalization of Convolutional Networks [article]

Nadav Cohen, Amnon Shashua
2014 arXiv   pre-print
Experiments demonstrate the capability of achieving state of the art accuracy with networks that are an order of magnitude smaller than comparable ConvNets.  ...  called MEX that realizes classical operators like ReLU and max pooling, but has additional capabilities that make SimNets a powerful generalization of ConvNets.  ...  Despite their success in recent years, ConvNets still fall short of reaching the holy grail of human-level visual recognition performance.  ... 
arXiv:1410.0781v3 fatcat:j7l3gfuejve6febosvvjghjms4

Automatic Shadow Detection and Removal from a Single Image

Salman H. Khan, Mohammed Bennamoun, Ferdous Sohel, Roberto Togneri
2016 IEEE Transactions on Pattern Analysis and Machine Intelligence  
Our proposed framework consistently performed better than the state-of-the-art on all major shadow databases collected under a variety of conditions.  ...  The model parameters are efficiently estimated using an iterative optimization procedure.  ...  ACKNOWLEDGMENTS The authors would like to thank the anonymous reviewers for their valuable comments and suggestions to improve the quality of the paper.  ... 
doi:10.1109/tpami.2015.2462355 pmid:27046489 fatcat:ti3xtpvl5zd7ppy2lsct6hihaq

PoseNet: A Convolutional Network for Real-Time 6-DOF Camera Relocalization

Alex Kendall, Matthew Grimes, Roberto Cipolla
2015 2015 IEEE International Conference on Computer Vision (ICCV)  
This is achieved using an efficient 23 layer deep convnet, demonstrating that convnets can be used to solve complicated out of image plane regression problems.  ...  Relocalization results for an input image (top), the predicted camera pose of a visual reconstruction (middle), shown again overlaid in red on the original image (bottom).  ...  (a) Relocalization with increasing levels of motion blur. The system is able to recognize the pose as high level features such as the contour outline still exist.  ... 
doi:10.1109/iccv.2015.336 dblp:conf/iccv/KendallGC15 fatcat:upc5umehj5a6dapdtbsegejhbm

Pavement Defect Segmentation in Orthoframes with a Pipeline of Three Convolutional Neural Networks

Roland Lõuk, Andri Riid, René Pihlak, Aleksei Tepljakov
2020 Algorithms  
The contribution also describes the complete procedure of working with pavement defects in an industrial setting, involving the workcycle of defect annotation, ConvNet training and validation.  ...  A novel neural network structure is considered, based on a pipeline of three ConvNets and endowed with the capacity for context awareness, which improves grid-based search for defects on orthoframes by  ...  However, it seems that adding context past the size of 336 × 336 is unnecessary. Different levels of context for a patch are visualized in Figure 16 .  ... 
doi:10.3390/a13080198 fatcat:dobs3i37sjgohildzctjseejgu

A Transformer-Based Siamese Network for Change Detection [article]

Wele Gedara Chaminda Bandara, Vishal M. Patel
2022 arXiv   pre-print
Different from recent CD frameworks, which are based on fully convolutional networks (ConvNets), the proposed method unifies hierarchically structured transformer encoder with Multi-Layer Perception (MLP  ...  This paper presents a transformer-based Siamese network architecture (abbreviated by ChangeFormer) for Change Detection (CD) from a pair of co-registered remote sensing images.  ...  illumination conditions.  ... 
arXiv:2201.01293v6 fatcat:qofs3zhc7feiff5ct7c3dnd4ny

Transforming medical imaging with Transformers? A comparative review of key properties, current progresses, and future perspectives [article]

Jun Li, Junyu Chen, Yucheng Tang, Bennett A. Landman, S. Kevin Zhou
2022 arXiv   pre-print
We conclude with discussions of future perspectives.  ...  After a brief introduction of the fundamentals of Transformers, especially in comparison with convolutional neural networks (CNNs), and highlighting key defining properties that characterize the Transformers  ...  Each path has a ConvNet backbone for generating high-level features from the input image/patches.  ... 
arXiv:2206.01136v1 fatcat:krji4fb2ivfulbu2biqx2ihsfa
« Previous Showing results 1 — 15 out of 659 results