Filters








521 Hits in 3.7 sec

Learning local feature descriptors with triplets and shallow convolutional neural networks

Vassileios Balntas, Edgar Riba, Daniel Ponsa, Krystian Mikolajczyk
2016 Procedings of the British Machine Vision Conference 2016   unpublished
It has recently been demonstrated that local feature descriptors based on convolutional neural networks (CNN) can significantly improve the matching performance.  ...  We compare our approach to recently introduced convolutional local feature descriptors, and demonstrate the advantages of the proposed methods in terms of performance and speed.  ...  Acknowledgements This work was supported by EPSRC project EP/N007743/1 and partially supported by the spanish project FireDMMI (TIN2014-56919-C3-2-R).  ... 
doi:10.5244/c.30.119 fatcat:4s232tgeebdxpmmxzzr2sp7la4

Multi-Scale Triplet CNN for Person Re-Identification

Jiawei Liu, Zheng-Jun Zha, QI Tian, Dong Liu, Ting Yao, Qiang Ling, Tao Mei
2016 Proceedings of the 2016 ACM on Multimedia Conference - MM '16  
In particular, we design a unified multi-scale network architecture consisting of both deep and shallow neural networks, towards learning robust and effective features for person re-identification under  ...  In this paper, we propose a multi-scale triplet convolutional neural network which captures visual appearance of a person at various scales.  ...  Qi Tian by ARO grants W911NF-15-1-0290 and Faculty Research Gift Awards by NEC Laboratories of America and Blippar.  ... 
doi:10.1145/2964284.2967209 dblp:conf/mm/LiuZTLYLM16 fatcat:dkozhjjbcjfc3iw4haelmya2wi

Deep Learning for Image Search and Retrieval in Large Remote Sensing Archives [article]

Gencer Sumbul, Jian Kang, Begüm Demir
2020 arXiv   pre-print
Initially, we analyze the limitations of the traditional CBIR systems that rely on the hand-crafted RS image descriptors.  ...  Then, we focus our attention on the advances in RS CBIR systems for which deep learning (DL) models are at the forefront.  ...  [21] Image pairs Random initialization Graph convolutional network Metric learning (supervised) Contrastive [22] Image triplets Pre-trained network weights Convolutional neural  ... 
arXiv:2004.01613v2 fatcat:d4fjt3vzybbbrejxzobaluqsoq

Hierarchical Metric Learning and Matching for 2D and 3D Geometric Correspondences [chapter]

Mohammed E. Fathy, Quoc-Huy Tran, M. Zeeshan Zia, Paul Vernaza, Manmohan Chandraker
2018 Lecture Notes in Computer Science  
We demonstrate that commonly used metric learning approaches do not optimally leverage the feature hierarchies learned in a Convolutional Neural Network (CNN), especially when applied to the task of geometric  ...  Recent advances in deep neural networks have enabled task-specific learned descriptors that outperform handcrafted descriptors on many problems.  ...  Recently, a few convolutional neural network (CNN) architectures [61, 16, 65, 58] have been proposed with the aim of learning strong geometric feature descriptors for matching images, and have yielded  ... 
doi:10.1007/978-3-030-01267-0_49 fatcat:jvhyw7ijc5e3vjd4rceuruzzsm

Hierarchical Metric Learning and Matching for 2D and 3D Geometric Correspondences [article]

Mohammed E. Fathy, Quoc-Huy Tran, M. Zeeshan Zia, Paul Vernaza, Manmohan Chandraker
2018 arXiv   pre-print
We demonstrate that commonly used metric learning approaches do not optimally leverage the feature hierarchies learned in a Convolutional Neural Network (CNN), especially when applied to the task of geometric  ...  Recent advances in deep neural networks have enabled task-specific learned descriptors that outperform hand-crafted descriptors on many problems.  ...  Recently, a few convolutional neural network (CNN) architectures [61, 16, 65, 58] have been proposed with the aim of learning strong geometric feature descriptors for matching images, and have yielded  ... 
arXiv:1803.07231v3 fatcat:cw6eb5kzkzggxebvuc37owm5sa

Content Based Image Retrieval Using Edge Based Feature Extraction In Deep Learning Algorithm

Siva Krishna Sajarao, M Pradeep
2019 Zenodo  
The convolutional neural network is having one of the layers is convolutional. It is used to extract the feature based on the different kernels to form the feature extraction map.  ...  Deep learning analysis is majorly concentrated on the different convolution layers. The convolutional neural network layer has been implemented on this process.  ...  CONVOLUTIONAL NEURAL NETWORK A significant CNN is retrained with comparability learning target work, considering triplets of material and irrelevant models procured from the totally related layers  ... 
doi:10.5281/zenodo.2606766 fatcat:fhvcf4n6b5hkxk47qo4u5fqgee

DF-SLAM: A Deep-Learning Enhanced Visual SLAM System based on Deep Local Features [article]

Rong Kang, Jieqi Shi, Xueming Li, Yang Liu, Xiao Liu
2019 arXiv   pre-print
We propose DF-SLAM system that uses deep local feature descriptors obtained by the neural network as a substitute for traditional hand-made features.  ...  Since we adopt a shallow network to extract local descriptors and remain others the same as original SLAM systems, our DF-SLAM can still run in real-time on GPU.  ...  Together with the metric learning layer, [24] uses triplet structure and achieves better performance. These achievements reveal the potential of triplet neural network.  ... 
arXiv:1901.07223v2 fatcat:bokvnvfmmvdhjanaid75jxxn5u

Local Deep Hashing Matching of Aerial Images Based on Relative Distance and Absolute Distance Constraints

Suting Chen, Xin Li, Yanyan Zhang, Rui Feng, Chuang Zhang
2017 Remote Sensing  
Secondly, a triplet network structure is proposed to mine the deep features of the patches of the local image, and the learned features are imported to the hash layer, thus obtaining the representation  ...  A triplet network was given in Vassileios Balntas etc. [19] and improved the loss function, finding that the matching time of the 128-dimension descriptor can be compared with the binary descriptor, such  ...  Due to the superiority in feature learning of deep convolutional neural networks and the superiority in retrieval calculating speed and storage space of hashing methods, deep convolution neural networks  ... 
doi:10.3390/rs9121244 fatcat:u3thue4hwvbszd62ycojil2dmu

TransLoc3D : Point Cloud based Large-scale Place Recognition using Adaptive Receptive Fields [article]

Tian-Xing Xu, Yuan-Chen Guo, Yu-Kun Lai, Song-Hai Zhang
2021 arXiv   pre-print
TransLoc3D consists of a 3D sparse convolutional module, an ARFM module, an external transformer network which aims to capture long range dependency and a NetVLAD layer.  ...  We also present a novel network architecture, named TransLoc3D, to obtain discriminative global descriptors of point clouds for the place recognition task.  ...  They propose to use a sparse convolutional neural network built on Feature Pyramid Network [22] to extract local descriptors.  ... 
arXiv:2105.11605v2 fatcat:s3zny5qnnbfipcfscpntxdr7ri

Survey on Deep Learning Techniques for Person Re-Identification Task [article]

Bahram Lavi, Mehdi Fatan Serj, Ihsan Ullah
2018 arXiv   pre-print
Among the systems, many researchers utilized deep neural networks (DNNs) because of their better performance and fast execution at test time.  ...  Intelligent video-surveillance is currently an active research field in computer vision and machine learning techniques.  ...  [55] proposed multi-scale triplet network copmpromised with one deep convolutional neural network and two shallow neural networks (i.e in order to produce less invariance and low-level appearance features  ... 
arXiv:1807.05284v3 fatcat:h4mtyitk3ncavfrfd7djlccp2y

Image-Based Geo-Localization Using Satellite Imagery [article]

Sixing Hu, Gim Hee Lee
2019 arXiv   pre-print
In particular, we show more extensive experimental results and analyses of the network architecture on our CVM-Net.  ...  To this end, in this paper we work on the extension of our earlier work on the Cross-View Matching Network (CVM-Net) for the ground-to-aerial image matching task since the traditional image descriptors  ...  Local feature extraction architectures We evaluate our CVM-Net-I with different convolutional neural network for local feature extractions.  ... 
arXiv:1903.00159v3 fatcat:jcsbr3dn5vh2zfmx3radzymuqq

Working hard to know your neighbor's margins: Local descriptor learning loss [article]

Anastasiya Mishchuk, Dmytro Mishkin, Filip Radenovic, Jiri Matas
2018 arXiv   pre-print
and deep convolution network architectures.  ...  We introduce a novel loss for learning local feature descriptors which is inspired by the Lowe's matching criterion for SIFT.  ...  and Economy, and the Province of Upper Austria in the frame of the COMET center, the CTU student grant SGS17/185/OHK3/3T/13, and the MSMT LL1303 ERC-CZ grant.  ... 
arXiv:1705.10872v4 fatcat:hdypqwvdmjccdhjhfktqrdsjyu

CVM-Net: Cross-View Matching Network for Image-Based Ground-to-Aerial Geo-Localization

Sixing Hu, Mengdan Feng, Rang M. H. Nguyen, Gim Hee Lee
2018 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition  
We first use the fully convolutional layers to extract local image features, which are then encoded into global image descriptors using the powerful NetVLAD.  ...  Specifically, our network is based on the Siamese architecture to do metric learning for the matching task.  ...  With the recent success of deep learning in many computer vision tasks, most of the existing works on cross-view image matching [49, 50, 46, 53] adopt the convolutional neural network (CNN) to learn  ... 
doi:10.1109/cvpr.2018.00758 dblp:conf/cvpr/HuFNL18 fatcat:hr5f6x64cvc2pixevcdfiulwzu

Fully Convolutional Network and Region Proposal for Instance Identification with Egocentric Vision

Maxime Portaz, Matthias Kohl, Georges Quenot, Jean-Pierre Chevallet
2017 2017 IEEE International Conference on Computer Vision Workshops (ICCVW)  
It is particularly suited for small datasets with low object variability. The proposed network can be trained end-to-end and produces an effective global descriptor as an image representation.  ...  This approach uses fully convolutional networks (FCN) to obtain region proposals without the need for an additional component in the network and training.  ...  The Neural Network model proposed is learned with a siamese network with three streams and a triplet loss [29] .  ... 
doi:10.1109/iccvw.2017.281 dblp:conf/iccvw/PortazKQC17 fatcat:33o4i5qiujfgtke37jgr7bam3u

Robust Angular Local Descriptor Learning [article]

Yanwu Xu, Mingming Gong, Tongliang Liu, Kayhan Batmanghelich, and Chaohui Wang
2019 arXiv   pre-print
In recent years, the learned local descriptors have outperformed handcrafted ones by a large margin, due to the powerful deep convolutional neural network architectures such as L2-Net [1] and triplet based  ...  loss function that gives smaller penalty to triplets with negative relative similarity.  ...  Different from the hand-crafted descriptors which extract low-level features such as gradients, the learned descriptors learn a convolutional neural network (CNN) from raw patches with ground-truth correspondences  ... 
arXiv:1901.07076v2 fatcat:5qx7txu4svddxoc2g7bououu3i
« Previous Showing results 1 — 15 out of 521 results