Filters








380,414 Hits in 5.0 sec

Learning to Assign Orientations to Feature Points [article]

Kwang Moo Yi, Yannick Verdie, Pascal Fua, Vincent Lepetit
2016 arXiv   pre-print
We show how to train a Convolutional Neural Network to assign a canonical orientation to feature points given an image patch centered on the feature point.  ...  To avoid the tedious and almost impossible task of finding a target orientation to learn, we propose to use Siamese networks which implicitly find the optimal orientations during training.  ...  The widelyused solution for assigning an orientation to a feature point is to use the dominant orientation of SIFT [25] .  ... 
arXiv:1511.04273v2 fatcat:44iokqd4afgqxk7dgohlhuuyrm

Learning to Assign Orientations to Feature Points

Kwang Moo Yi, Yannick Verdie, Pascal Fua, Vincent Lepetit
2016 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
Compared Methods To keep the maximum number of features points to 1000, we sort the detected feature points according to their respective response scores and keep the best 1000.  ...  Other parameters are set to default values. which is the same as what SIFT uses.  ...  We use the pre-learned model learned with the liberty dataset from [34], as the other two datasets are partially included in our test set. • LIOP [42]: VLFeat library -http://www.vlfeat.org/ Default parameters  ... 
doi:10.1109/cvpr.2016.19 dblp:conf/cvpr/YiVFL16 fatcat:gyt3cis4dncbnjfwb4o3v3kxkm

Oriented RepPoints for Aerial Object Detection [article]

Wentong Li, Yijie Chen, Kaixuan Hu, Jianke Zhu
2022 arXiv   pre-print
Moreover, we propose an effective quality assessment and sample assignment scheme for adaptive points learning toward choosing the representative oriented reppoints samples during training, which is able  ...  A spatial constraint is introduced to penalize the outlier points for roust adaptive learning.  ...  Adaptive Points Assessment and Assignment Due to the lack of direct supervision, learning highquality points is essential to capture the geometric features adaptively for densely-packed and arbitrarily-oriented  ... 
arXiv:2105.11111v4 fatcat:rufkrf4oenhj5ad5yisgrnbeau

DEEP LEARNING BASED FEATURE MATCHING AND ITS APPLICATION IN IMAGE ORIENTATION

L. Chen, F. Rottensteiner, C. Heipke
2020 ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences  
Affine shape estimation, orientation assignment and feature description algorithms based on detected hand crafted features have shown to be error prone.  ...  In this paper, affine shape estimation, orientation assignment and description of local features is achieved through deep learning.  ...  ACKNOWLEDGEMENTS The authors would like to thank NVIDIA Corp. for donating the GPU used in this research through its GPU grant program.  ... 
doi:10.5194/isprs-annals-v-2-2020-25-2020 fatcat:7nplhqf7ubavngqysdukkygtzi

Learnable Stroke Models for Example-based Portrait Painting

Tinghuai Wang, John Collomosse, Andrew Hunter, Darryl Greig
2013 Procedings of the British Machine Vision Conference 2013  
Style models local to facial features are learned using a semantic segmentation of the input face image, driven by a combination of an Active Shape Model and Graph-cut.  ...  Given a training pair -a source image and painting of that image -a non-parametric model of style is learned by observing the geometry and tone of brush strokes local to image features.  ...  the stroke property set P given features F present in the image local to these points.  ... 
doi:10.5244/c.27.36 dblp:conf/bmvc/WangCHG13 fatcat:ikv6eyumzbeglct2z7kqamkjlu

SphereVLAD++: Attention-based and Signal-enhanced Viewpoint Invariant Descriptor [article]

Shiqi Zhao, Peng Yin, Ge Yi, Sebastian Scherer
2022 arXiv   pre-print
SphereVLAD++ projects the point cloud on the spherical perspective for each unique area and captures the contextual connections between local features and their dependencies with global 3D geometry distribution  ...  the distinguishable feature extraction ability.  ...  Moreover recently, SOE-Net [7] modify PointNet by adding orientational feature embeddings, which learn local geometry from eight spatial orientations to reduce the sensitive to orientation variances.  ... 
arXiv:2207.02958v1 fatcat:w2lwqdk2dbhlfk53xhufqqpc3y

Learning Covariant Feature Detectors [article]

Karel Lenc, Andrea Vedaldi
2016 arXiv   pre-print
We then derive a covariance constraint that can be used to automatically learn which visual structures provide stable anchors for local feature detection.  ...  In this paper, we propose the first fully general formulation for learning local covariant feature detectors.  ...  Acknowledgements We would like to thank ERC 677195-IDIU for supporting this research.  ... 
arXiv:1605.01224v2 fatcat:o2askz32mza4lneiqzexogwnoy

Latent fingerprint minutia extraction using fully convolutional network [article]

Yao Tang, Fei Gao, Jufu Feng
2017 arXiv   pre-print
As the limitation of traditional handcrafted features, a fully convolutional network (FCN) is utilized to learn features directly from data to overcome complex background noises.  ...  Then small regions centering at these minutia points are entered into a convolutional neural network (CNN) to reclassify these minutiae and calculate their orientations.  ...  Minutiae extraction is seen as a point detection problem and the features are learned from data automatically to adapt to complex background. 2.  ... 
arXiv:1609.09850v2 fatcat:vtsiqoxw7bdwdnons6t2s57k6u

Orientation-boosted Voxel Nets for 3D Object Recognition [article]

Nima Sedaghat, Mohammadreza Zolfaghari, Ehsan Amiri, Thomas Brox
2017 arXiv   pre-print
In this paper, we show that the object orientation plays an important role in 3D recognition. More specifically, we argue that objects induce different features in the network under rotation.  ...  Thus, we approach the category-level classification task as a multi-task problem, in which the network is trained to predict the pose of the object in addition to the class label as a parallel task.  ...  Thus we converted the 3D mesh grids of Modelnet40 to point-clouds by assigning uniformly distributed points to object faces.  ... 
arXiv:1604.03351v2 fatcat:im72dwdldje6tagvtk2n3koina

Local Feature Detectors, Descriptors, and Image Representations: A Survey [article]

Yusuke Uchida
2016 arXiv   pre-print
In addition, recent deep learning-based approaches for image retrieval are briefly reviewed.  ...  %All of the local feature-based image retrieval system involves two important processes: local feature extraction and image representation.  ...  In [138] , a learning scheme based on CNN is introduced to estimate a canonical orientation for local features.  ... 
arXiv:1607.08368v1 fatcat:g3qm64yonvcd5fe5sw4zbn3mo4

A Comparison of CNN and Classic Features for Image Retrieval [article]

Umut Özaydın, Theodoros Georgiou, Michael Lew
2019 arXiv   pre-print
The results show that each type of features are best in different contexts.  ...  Feature detectors and descriptors have been successfully used for various computer vision tasks, such as video object tracking and content-based image retrieval.  ...  Moreover, there are several features that a method may or may not have, such as orientation assignment and whether it provides a description method.  ... 
arXiv:1908.09300v1 fatcat:l2r4guozwbhktffllc3l67jrpu

PointSIFT: A SIFT-like Network Module for 3D Point Cloud Semantic Segmentation [article]

Mingyang Jiang, Yiran Wu, Tianqi Zhao, Zelin Zhao, Cewu Lu
2018 arXiv   pre-print
Recently, 3D understanding research sheds light on extracting features from point cloud directly, which requires effective shape pattern description of point clouds.  ...  Specifically, an orientation-encoding unit is designed to describe eight crucial orientations, and multi-scale representation is achieved by stacking several orientation-encoding units.  ...  The downsampling is implemented by finding N centroids with farthest point sampling, assigning points to centroids and then calculate embedding of centroids by feeding feature of assigned points through  ... 
arXiv:1807.00652v2 fatcat:5bu57wkz5ne4tkohdd65k2xviq

Learning Sensor Interdependencies for IMU-to-Segment Assignment

Tomoya Kaichi, Tsubasa Maruyama, Mitsunori Tada, Hideo Saito
2021 IEEE Access  
We address this IMU-to-segment assignment problem and propose a novel end-to-end learning model that incorporates a global feature generation module and an attention-based mechanism.  ...  The former extracts the feature representing the motion of all attached IMUs, and the latter enables the model to learn the dependency relationships between the IMUs.  ...  Pointnet [23] is the pioneering work in applying neural networks to learn over general point sets.  ... 
doi:10.1109/access.2021.3105801 fatcat:iths5olcrfhjfkuf2hhkifokwy

Using Inertial Data to Enhance Image Segmentation - Knowing Camera Orientation Can Improve Segmentation of Outdoor Scenes

Osian Haines, David Bull, J. F. Burn
2015 Proceedings of the 10th International Conference on Computer Vision Theory and Applications  
The method is applied to segmentation using both points and lines, and we also show that combining points with lines further improves accuracy.  ...  We show that orientation information is useful in conjunction with typical image-based features, and that fusing the two results in substantially better classification accuracy than either alone -we observed  ...  The authors would like to thank Austin Gregg-Smith for advice on hardware and graphics, and Dr David Hanwell for help with maths and text.  ... 
doi:10.5220/0005274000210032 dblp:conf/visapp/HainesBB15 fatcat:io6zbm57tbepvemddra5gv2tfi

Object Recognition using Still Image

Deepika T.V, Stafford Michahial, Dr M Shiva kumar
2016 IJARCCE  
It is also robust to small amount of out-of-plane rotation and conclusion.  ...  This paper presents an algorithm for detecting a specific object based on finding point correspondences between the reference and the target image.  ...  For such time-critical applications, point feature matching is an attractive solution because new objects can be easily learned online, in contrast to statistical-learning techniques that require many  ... 
doi:10.17148/ijarcce.2016.51242 fatcat:zxhnschkczfzrcltjirlbnbdpu
« Previous Showing results 1 — 15 out of 380,414 results