4,419 Hits in 4.4 sec

Indexing and searching 100M images with map-reduce

Diana Moise, Denis Shestakov, Gylfi Gudmundsson, Laurent Amsaleg
2013 Proceedings of the 3rd ACM conference on International conference on multimedia retrieval - ICMR '13  
High-Dimensional Indexing, Map-Reduce, Hadoop. 2. Index search with Map-Reduce. For some applications, throughput is way more important than is the response time of individual queries.  ...  We share our experience of adapting a clustering-based high-dimensional indexing algorithm to the Map-Reduce model, and of testing it at large scale with Hadoop as we index 30 billion SIFT descriptors.  ...  Hadoop is now the de-facto reference Map-Reduce implementation. . #3: 100M images, 30.2B Descs., 4TB Figure 1 : 1 . #4: Time and Quality, 100M images Searching the entire batch took 1,623 sec. on average  ... 
doi:10.1145/2461466.2461470 dblp:conf/mir/MoiseSGA13 fatcat:jiaj456w5zgujaqeq22ualewty

Distributed Kd-Trees for Ultra Large Scale Object Recognition

Mohamed Aly, Mario Munich, Pietro Perona
2011 Procedings of the British Machine Vision Conference 2011  
We show experiments with up to 100 million images running on 2048 machines, with run time of a fraction of a second for each query image.  ...  It can run on thousands of machines, and provides orders of magnitude more throughput than the state-of-the-art, with better recognition performance.  ...  We would also like to thank James Philbin, Hartwig Adam, and Hartmut Neven for their valuable help.  ... 
doi:10.5244/c.25.40 dblp:conf/bmvc/AlyMP11 fatcat:ej6xbbrxondybhvfo5qzprzkby

Scalable high-dimensional indexing and searching with Hadoop [article]

Denis Shestakov, Diana Moise
2015 arXiv   pre-print
Besides its scalability, the proposed scheme not only provides good search quality, but also achieves a stable throughput of around 210ms per image when searching a 100M image collection.  ...  In this paper, we show how high-dimensional indexing and searching methods can be used on scientific grid environments and present a scalable workflow for indexing and searching over 30 billion SIFT descriptors  ...  First, we evaluated the quality of search results by searching the query batch across the full dataset (100M images, 4TB size) and 20% of it (20M, 1TB size).  ... 
arXiv:1501.02398v2 fatcat:ut5aipv4arejtaan4kh2ywv4fe

Scalable high-dimensional indexing with Hadoop

Denis Shestakov, Diana Moise, Gylfi Gudmundsson, Laurent Amsaleg
2013 2013 11th International Workshop on Content-Based Multimedia Indexing (CBMI)  
In this paper, we show how high-dimensional indexing methods can be used on scientific grid environments and present a scalable workflow for indexing and searching over 30 billion SIFT descriptors using  ...  While high-dimensional search-by-similarity techniques reached their maturity and in overall provide good performance, most of them are unable to cope with very large multimedia collections.  ...  Besides its scalability benefit, the proposed scheme not only achieves good search quality but also allows to search 100M image collection with a stable throughput of around 210ms per image.  ... 
doi:10.1109/cbmi.2013.6576584 dblp:conf/cbmi/ShestakovMGA13 fatcat:6khy2kh3rnbzfbnwjyryumcfdu

Aggregating Local Image Descriptors into Compact Codes

H. Jegou, F. Perronnin, M. Douze, J. Sanchez, P. Perez, C. Schmid
2012 IEEE Transactions on Pattern Analysis and Machine Intelligence  
This paper addresses the problem of large-scale image search. Three constraints have to be taken into account: search accuracy, efficiency, and memory usage.  ...  The evaluation shows that the image representation can be reduced to a few dozen bytes while preserving high accuracy. Searching a 100 million image dataset takes about 250 ms on one processor core.  ...  ACKNOWLEDGEMENTS We thank the company Exalead for providing the corpus of 100M images.  ... 
doi:10.1109/tpami.2011.235 pmid:22156101 fatcat:vdo4fohvnne35fp5cguxei3uza

Generation of DEM by radargrammetric techniques

Franck Fayard, Stephane Meric, Eric Pottier
2010 2010 IEEE International Geoscience and Remote Sensing Symposium  
Working area image, SRTM DEM and generated DEM pixels with altimetric error Pyramidal step < 20m < 50m < 100m < 200mTable 1.  ...  Finally, the epipolar geometry considerably reduces the size of the search area in both range and azimuth directions. (a) Epipolar search area (b) Pyramidal scheme with growing resolutions Fig. 3 .  ... 
doi:10.1109/igarss.2010.5654111 dblp:conf/igarss/FayardMP10a fatcat:5rsse663vbaejftpr7xqzamhuq

Polysemous codes [article]

Matthijs Douze, Hervé Jégou, Florent Perronnin
2016 arXiv   pre-print
Last but not least, our approach allows the approximate computation of the k-NN graph associated with the Yahoo Flickr Creative Commons 100M, described by CNN image descriptors, in less than 8 hours on  ...  The method is complementary with a coarse partitioning of the feature space such as the inverted multi-index.  ...  This yields 4096-dimensional image descriptors. Prior to indexing we reduce these descriptors to 256D with PCA and subsequently apply a random rotation [9, 29] .  ... 
arXiv:1609.01882v2 fatcat:xxrag423znd6nfnjrg3o7j3j7m

Stereo-vision-based terrain mapping for off-road autonomous navigation

Arturo L. Rankin, Andres Huertas, Larry H. Matthies, Grant R. Gerhart, Douglas W. Gage, Charles M. Shoemaker
2009 Unmanned Systems Technology XI  
A compact terrain map is built from each frame of stereo images.  ...  In this paper, we summarize the terrain mapping capabilities that JPL has implemented during several UGV programs over the last decade and discuss some challenges to building terrain maps with stereo range  ...  by ARL under the Demo III and RCTA programs, through agreements with the National Aeronautics and Space Administration (NASA).  ... 
doi:10.1117/12.819099 fatcat:2fyuqahzofbk5mphsofyvlk3ve

Efficient Large-Scale Similarity Search Using Matrix Factorization

Ahmet Iscen, Michael Rabbat, Teddy Furon
2016 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
Our goal is to reduce the number of vector operations and memory for performing a search without sacrificing accuracy of the returned images.  ...  Experiments with standard image search benchmarks, including the Ya-hoo100M dataset comprising 100 million images, show that our method gives comparable (and sometimes superior) accuracy compared to exhaustive  ...  Acknowledgments This work was supported, in part, by the Natural Sciences and Engineering Research Council of Canada through grant RGPAS 429296-12, and by the French National Project IDFRAud (ANR-14-CE28  ... 
doi:10.1109/cvpr.2016.228 dblp:conf/cvpr/IscenRF16 fatcat:zzglnrliurdyxgogf6b5p6pf7q

Video-based localization without 3D mapping for the visually impaired

Jason J. Liu, Cody Phillips, Kostas Daniilidis
2010 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops  
We assume that a video sequence has been acquired and that keyframes have been registered with respect to 2D positions and orientations.  ...  We present a novel approach to localization by using search space reduction on global features, then HMM based position prediction and estimation on local features.  ...  Database registration: Extract GIST and SURF features and register positions. 3. Online Localization: for each new query image: (a) GIST matching: calculate reduced search space.  ... 
doi:10.1109/cvprw.2010.5543581 dblp:conf/cvpr/LiuPD10 fatcat:3vdqypjb4bawtolp3hedvxwwti

Towards Engineering a Web-Scale Multimedia Service

Gylfi Þór Guðmundsson, Laurent Amsaleg, Björn Þór Jónsson, Michael J. Franklin
2017 Proceedings of the 8th ACM on Multimedia Systems Conference - MMSys'17  
Map-Reduce and Hadoop Map-Reduce is from Google [5] and exploits data independence through the Map and Reduce user-level functions.  ...  The largest experiments on Hadoop indexed and searched around 100M images, or about 30 billion SIFT descriptors, using a cluster of 100+ machines [17] .  ... 
doi:10.1145/3083187.3083200 dblp:conf/mmsys/Gudmundsson17 fatcat:5cxbxujlabev7ji6x5pyt3knre

YFCC100M HybridNet fc6 Deep Features for Content-Based Image Retrieval

Giuseppe Amato, Fabrizio Falchi, Claudio Gennaro, Fausto Rabitti
2016 Proceedings of the 2016 ACM Workshop on Multimedia COMMONS - MMCommons '16  
This set of results is ground truth for evaluating Content-Based Image Retrieval (CBIR) systems that use approximate similarity search methods for efficient and scalable indexing.  ...  Moreover, we present experimental results obtained indexing this corpus with two distinct approaches: the Metric Inverted File and the Lucene Quantization.  ...  Performance Evaluation of the indexes The two CBIR systems use approximate similarity search indexes, i.e., indexes that introduce some errors in the search results, to provide very high query execution  ... 
doi:10.1145/2983554.2983557 fatcat:ayqkkrvufbgmxoicy35iogneju

CNN Architectures for Large-Scale Audio Classification [article]

Shawn Hershey, Sourish Chaudhuri, Daniel P. W. Ellis, Jort F. Gemmeke, Aren Jansen, R. Channing Moore, Manoj Plakal, Devin Platt, Rif A. Saurous, Bryan Seybold, Malcolm Slaney, Ron J. Weiss, Kevin Wilson
2017 arXiv   pre-print
Convolutional Neural Networks (CNNs) have proven very effective in image classification and show promise for audio.  ...  We investigate varying the size of both training set and label vocabulary, finding that analogs of the CNNs used in image classification do well on our audio classification task, and larger training and  ...  ACKNOWLEDGEMENTS The authors would like to thank George Toderici and Marvin Ritter, both with Google, for their very valuable feedback.  ... 
arXiv:1609.09430v2 fatcat:y7ljzpflkvgmhjdiae2o5bhqn4

Fast and robust duplicate image detection on the web

Etienne Gadeski, Hervé Le Borgne, Adrian Popescu
2016 Multimedia tools and applications  
However, in the context considered, it requires not only an efficient indexing and searching algorithm but also to be fast to compute the image description, since the total time of description and searching  ...  We evaluate our method on two different datasets with the use of different sets of distractor images, leading to large-scale image collections (up to 100 million images).  ...  Conclusions We introduce a fast and robust image description which is well suited for indexing and searching through image data streams.  ... 
doi:10.1007/s11042-016-3619-4 fatcat:nvqt6h2zpja3tpzqq2b7xnlo6e


C. L. Mallmann, A. F. Zaninni, W. Pereira Filho
2020 The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences  
In this study, search to evaluate the vegetation index potential by using multispectral images acquired by UAV as a support on detection and monitoring of invasive plants in Quarta Colônia State Park located  ...  A sampling area with a density of invasive plants above 80% was evaluated, with predominance of Psidium guajava and Ligustrum lucidum, generating a large data set from the extracted indexes.  ...  ACKNOWLEDGEMENTS Acknowledgements of support for the project from the State Secretary of environment and infrastructure of RS, Brazil and Group CPFL Energy.  ... 
doi:10.5194/isprs-archives-xlii-3-w12-2020-521-2020 fatcat:ugsi65zs7nht7gjvlivdy2tqqa
« Previous Showing results 1 — 15 out of 4,419 results