Filters








1,841 Hits in 1.4 sec

Predicting Matchability

Wilfried Hartmann, Michal Havlena, Konrad Schindler
2014 2014 IEEE Conference on Computer Vision and Pattern Recognition  
We show that one can in fact learn to predict which descriptors are matchable, and thus reduce the number of interest points significantly without losing too many matches.  ...  Moreover, we embed the prediction in a state-of-theart Structure-from-Motion pipeline and demonstrate that it also outperforms other selection methods at system level.  ...  Next, we quantify the speed-up achieved through the matchability prediction.  ... 
doi:10.1109/cvpr.2014.9 dblp:conf/cvpr/HartmannHS14 fatcat:y5mt6y4emraqjhtkmopz2rxxkq

Matchability Prediction for Full-Search Template Matching Algorithms

Adrian Penate-Sanchez, Lorenzo Porzi, Francesc Moreno-Noguer
2015 2015 International Conference on 3D Vision  
By using deep learning descriptions of patches we are able to predict matchability over the whole image quite reliably.  ...  In this paper we alleviate the computational load of these algorithms by proposing an efficient approach for predicting the matchability of a template, before it is actually performed.  ...  applied to the source images, selecting fifty templates corresponding to the local maxima in the matchability maps with the highest predicted matching probability.  ... 
doi:10.1109/3dv.2015.47 dblp:conf/3dim/SanchezPM15 fatcat:5ohexg2innfmdlkd3vod2sc3te

Learning Dense Correspondence via 3D-guided Cycle Consistency [article]

Tinghui Zhou, Philipp Krähenbühl, Mathieu Aubry, Qixing Huang, Alexei A. Efros
2016 arXiv   pre-print
We exploit this consistency as a supervisory signal to train a convolutional neural network to predict cross-instance correspondences between pairs of images depicting objects of the same category.  ...  We use ground-truth synthetic-to-synthetic correspondences, provided by the rendering engine, to train a ConvNet to predict synthetic-to-real, real-to-real and real-to-synthetic correspondences that are  ...  Matchability prediction We evaluate the performance of matchability prediction using the PASCAL-Part dataset [8] , which provides humanannotated part segment labeling 2 .  ... 
arXiv:1604.05383v1 fatcat:43ck5wohhjfjhetqlfcz7puntm

Learning Dense Correspondence via 3D-Guided Cycle Consistency

Tinghui Zhou, Philipp Krahenbuhl, Mathieu Aubry, Qixing Huang, Alexei A. Efros
2016 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
We exploit this consistency as a supervisory signal to train a convolutional neural network to predict cross-instance correspondences between pairs of images depicting objects of the same category.  ...  We use ground-truth synthetic-to-synthetic correspondences, provided by the rendering engine, to train a ConvNet to predict synthetic-to-real, real-to-real and realto-synthetic correspondences that are  ...  Ours final Matchability prediction We evaluate the performance of matchability prediction using the PASCAL-Part dataset [8] , which provides humanannotated part segment labeling 2 .  ... 
doi:10.1109/cvpr.2016.20 dblp:conf/cvpr/ZhouKAHE16 fatcat:miasm7gltzhmxh7c6xjmhkku5q

Match or No Match: Keypoint Filtering based on Matching Probability

Alexandra I. Papadaki, Ronny Hansch
2020 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)  
We introduce an algorithm that filters detected keypoints before the matching is even attempted, by predicting the probability of each point to be successfully matched.  ...  proposed method decreases the computational cost of a subsequent keypoint matching and 3D reconstruction, by correctly filtering 50% of the points that wouldn't be matched while preserving 73% of the matchable  ...  In this step, a keypoint classification is performed to predict and preserve the matchable keypoints.  ... 
doi:10.1109/cvprw50498.2020.00515 dblp:conf/cvpr/PapadakiH20 fatcat:fbfokicv3rbc7kwhcp235cwidq

Analyzing and revising data integration schemas to improve their matchability

Xiaoyong Chai, Mayssam Sayyadian, AnHai Doan, Arnon Rosenthal, Len Seligman
2008 Proceedings of the VLDB Endowment  
Next, mSeer uses this score to generate a matchability report that identifies the problems in matching S.  ...  Given a mediated schema S and a matching tool M , mSeer first computes a matchability score that quantifies how well S can be matched against using M .  ...  Predict a Wrong Match: s = t, but M predicts s = t ′ . The mistake in this case is two-fold. First, M fails to predict the correct match s = t, which implies sim(s, t) < ǫ.  ... 
doi:10.14778/1453856.1453940 fatcat:g6ocssnw2jfvvejhp472p4lt7q

Learning Dense Facial Correspondences in Unconstrained Images [article]

Ronald Yu, Shunsuke Saito, Haoxiang Li, Duygu Ceylan, Hao Li
2017 arXiv   pre-print
Our network learns a per-pixel flow and a matchability mask between 2D input photographs of a person and the projection of a textured 3D face model.  ...  We also predict a matchability mask which indicates which correspondences are valid or not.  ...  Both 2D flow and the matchability are predicted by a convolutional neural network (Section 3.3).  ... 
arXiv:1709.00536v1 fatcat:yvqh5ntb7nbpzd2yexo232rhea

Matching Features Correctly through Semantic Understanding

Nikolay Kobyshev, Hayko Riemenschneider, Luc Van Gool
2014 2014 2nd International Conference on 3D Vision  
Second, we propose to learn correct matchability of descriptors from these semantic contexts.  ...  Every method has a confidence that determines how many points it will predict to be correctly matchable.  ...  First, instead of trying to predict if a feature will be matchable or not, we are aiming for leaving only the features that are predicted to be matched correctly.  ... 
doi:10.1109/3dv.2014.15 dblp:conf/3dim/KobyshevRG14 fatcat:trfaynb7w5d5foogvvttm46ag4

Optimizing color consistency in photo collections

Yoav HaCohen, Eli Shechtman, Dan B. Goldman, Dani Lischinski
2013 ACM Transactions on Graphics  
Below we describe our matchability classifier and link prediction strategy in more detail.  ...  as matchable}| ), on six test albums with threshold 0 (this is the SVM's threshold that we use in the first phase of the link prediction algorithm).  ... 
doi:10.1145/2461912.2461997 fatcat:osm3cvjzgrhj3kenxd4qw2n7iq

Dangling-Aware Entity Alignment with Mixed High-Order Proximities [article]

Juncheng Liu, Zequn Sun, Bryan Hooi, Yiwei Wang, Dayiheng Liu, Baosong Yang, Xiaokui Xiao, Muhao Chen
2022 arXiv   pre-print
Extensive experiments with two evaluation settings shows that our framework more precisely detects dangling entities, and better aligns matchable entities.  ...  for the predicted matchable entities.  ...  Arjovsky et al. (2017) Table 5 : Some dangling source entities wrongly predicted as matchable by the previous method, while MHP predicts them as dangling with high probabilities. Cls.  ... 
arXiv:2205.02406v1 fatcat:wqe6gvlhpbaupmf2yqmvppgvey

Learning Stereo Matchability in Disparity Regression Networks [article]

Jingyang Zhang, Yao Yao, Zixin Luo, Shiwei Li, Tianwei Shen, Tian Fang, Long Quan
2020 arXiv   pre-print
Finally, a matchability-aware disparity refinement is introduced to improve the depth inference in weakly matchable regions.  ...  In this paper, we address this challenge by proposing a stereo matching network that considers pixel-wise matchability.  ...  Joint Disparity and Matchability Learning Assuming that the observed disparity value d gt follows the Laplacian distribution [10] , we predict a mean disparity valued and a scale factorb (sometimes referred  ... 
arXiv:2008.04800v1 fatcat:ym4hxth6cbdmvf3qj5niotqr6e

RANSAC-Flow: generic two-stage image alignment [article]

Xi Shen, François Darmon, Alexei A. Efros, Mathieu Aubry
2020 arXiv   pre-print
s (x, y)M s→t (x , y ) (2) where M s→t is the matchability predicted from source to target and M t→s the one predicted from target to source.  ...  Stage 2: given two coarsely aligned images, our self-supervised fine flow network generates flow predictions in the matchable region.  ... 
arXiv:2004.01526v2 fatcat:ltpp6c4gcnfmdo3pxdt2ulqqp4

Filtration and Distillation: Enhancing Region Attention for Fine-Grained Visual Categorization

Chuanbin Liu, Hongtao Xie, Zheng-Jun Zha, Lingfeng Ma, Lingyun Yu, Yongdong Zhang
2020 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
Firstly, a Filtration Learning (FL) method is put forward for discriminative part regions proposing based on the matchability between proposing and predicting.  ...  Specifically, we utilize the proposing-predicting matchability as the performance metric of Region Proposal Network (RPN), thus enable a direct optimization of RPN to filtrate most discriminative regions  ...  In this paper, we propose a filtration learning method that utilizes the matchability between proposing and predicting to optimize region proposal.  ... 
doi:10.1609/aaai.v34i07.6822 fatcat:bsypoxkne5fpfcm26kot7nlxq4

Alignment Strength and Correlation for Graphs [article]

Donniell E. Fishkind, Lingyao Meng, Ao Sun, Carey E. Priebe, Vince Lyzinski
2020 arXiv   pre-print
in matchability.  ...  Within broad families of the random graph parameter settings, we illustrate that exact graph matching runtime and also matchability are both functions of ρ_T, with thresholding behavior starkly illustrated  ...  In particular, the empirical matchability demonstrations in this paper are not predictable from the previously known matchability asymptotics.  ... 
arXiv:1808.08502v4 fatcat:6lvegm4ydzfjvdeuukl2uuyd5e

Alignment strength and correlation for graphs

Donniell E. Fishkind, Lingyao Meng, Ao Sun, Carey E. Priebe, Vince Lyzinski
2019 Pattern Recognition Letters  
in matchability.  ...  Within broad families of the random graph parameter settings, we illustrate that exact graph matching runtime and also matchability are both functions of ϱ T , with thresholding behavior starkly illustrated  ...  In particular, the empirical matchability demonstrations in this paper are not predictable from the previously known matchability asymptotics.  ... 
doi:10.1016/j.patrec.2019.05.008 fatcat:d3h7sjnasvamtiqeamzp22fpde
« Previous Showing results 1 — 15 out of 1,841 results