22 Hits in 6.6 sec

Efficient Object Category Recognition Using Classemes [chapter]

Lorenzo Torresani, Martin Szummer, Andrew Fitzgibbon
2010 Lecture Notes in Computer Science  
We introduce a new descriptor for images which allows the construction of efficient and compact classifiers with good accuracy on object category recognition.  ...  , and allows these queries to be for novel categories.  ...  Large-scale recognition benefits from a compact descriptor for each image, for example allowing databases to be stored in memory rather than on disk.  ... 
doi:10.1007/978-3-642-15549-9_56 fatcat:vg5jc46ehbdmjdodlornaaaiou

Classemes and Other Classifier-Based Features for Efficient Object Categorization

Alessandro Bergamo, Lorenzo Torresani
2014 IEEE Transactions on Pattern Analysis and Machine Intelligence  
This paper describes compact image descriptors enabling accurate object categorization with linear classification models, which offer the advantage of being efficient to both train and test.  ...  specifically tuned for a predefined set of test classes.  ...  ACKNOWLEDGMENT We are grateful to Andrew Fitzgibbon for his contribution in the design of classemes and PICODES. We thank Martin Szummer for discussions and Chen Fang for programming help.  ... 
doi:10.1109/tpami.2014.2313111 pmid:26352630 fatcat:5qyanhgnpjda5ep3grr457lhja

Meta-class features for large-scale object categorization on a budget

A. Bergamo, L. Torresani
2012 2012 IEEE Conference on Computer Vision and Pattern Recognition  
In this paper we introduce a novel image descriptor enabling accurate object categorization even with linear models.  ...  By using "learnability" of the meta-classes as criterion for feature generation, we obtain a set of attributes that encode general visual properties shared by multiple object classes and that are effective  ...  Acknowledgements We are grateful to Jia Deng and Florent Perronnin for answering questions about their systems. This research was funded in part by NSF CAREER award IIS-0952943.  ... 
doi:10.1109/cvpr.2012.6248040 dblp:conf/cvpr/BergamoT12 fatcat:5dq7i7h3knfqbkq2eurniufd6m

Scalable object-class retrieval with approximate and top-k ranking

Mohammad Rastegari, Chen Fang, Lorenzo Torresani
2011 2011 International Conference on Computer Vision  
Experimental results show that our algorithms for object-class retrieval can search a 10 million database in just a couple of seconds and produce categorization accuracy comparable to the best known class-recognition  ...  In this paper we address the problem of object-class retrieval in large image data sets: given a small set of training examples defining a visual category, the objective is to efficiently retrieve images  ...  This research was funded in part by Microsoft and NSF CAREER award IIS-0952943.  ... 
doi:10.1109/iccv.2011.6126556 dblp:conf/iccv/RastegariFT11 fatcat:uaqkjfrstfh4tenlyr2q3yq7ua

What can pictures tell us about web pages?

Sergio Rodriguez-Vaamonde, Lorenzo Torresani, Andrew Fitzgibbon
2013 Proceedings of the 36th international ACM SIGIR conference on Research and development in information retrieval - SIGIR '13  
Traditional Web search engines do not use the images in the HTML pages to find relevant documents for a given query.  ...  We present a Web-scalable system that exploits a pure text-based search engine to find an initial set of candidate documents for a given query.  ...  We address the first requirement by utilizing a compact attribute-based image descriptor-the classeme vector [6] -which has been shown to yield accurate object recognition even with simple linear classifiers  ... 
doi:10.1145/2484028.2484144 dblp:conf/sigir/Rodriguez-VaamondeTF13 fatcat:7int6otckfdgrhdinreia7l5ie

What Can Pictures Tell Us About Web Pages? Improving Document Search Using Images

Sergio Rodriguez-Vaamonde, Lorenzo Torresani, Andrew W. Fitzgibbon
2015 IEEE Transactions on Pattern Analysis and Machine Intelligence  
Traditional Web search engines do not use the images in the HTML pages to find relevant documents for a given query.  ...  We present a Web-scalable system that exploits a pure text-based search engine to find an initial set of candidate documents for a given query.  ...  We address the first requirement by utilizing a compact attribute-based image descriptor-the classeme vector [6] -which has been shown to yield accurate object recognition even with simple linear classifiers  ... 
doi:10.1109/tpami.2014.2366761 pmid:26357348 fatcat:aqen52qxbbf63cos3i6wlbc65i

Efficient aspect object models using pre-trained convolutional neural networks

Eric Wilkinson, Takeshi Takahashi
2015 2015 IEEE-RAS 15th International Conference on Humanoid Robots (Humanoids)  
In this paper, we maintain an object model database consisting of aspect and class descriptors computed from images of target objects at varying view points.  ...  We investigate the effectiveness of different dimensionality reduction techniques for key generation on query efficiency and accuracy.  ...  To reduce query times and handle scalability we use a Ball Tree index structure that efficiently performs nearest neighbor searches even for high dimensional data [9] .  ... 
doi:10.1109/humanoids.2015.7363556 dblp:conf/humanoids/WilkinsonT15 fatcat:id7fazej4jaupbgyhw2mitbco4

Harvesting Mid-level Visual Concepts from Large-Scale Internet Images

Quannan Li, Jiajun Wu, Zhuowen Tu
2013 2013 IEEE Conference on Computer Vision and Pattern Recognition  
Here, we take the advantage of having massive wellorganized Google and Bing image data; visual concepts (around 14, 000) are automatically exploited from images using word-based queries.  ...  In this paper, we propose a fully automatic algorithm which harvests visual concepts from a large number of Internet images (more than a quarter of a million) using text-based queries.  ...  and object recognition.  ... 
doi:10.1109/cvpr.2013.115 dblp:conf/cvpr/LiWT13 fatcat:2vx5jb7xdvewbmml5bwsp33fzi

Measuring Image Distances via Embedding in a Semantic Manifold [chapter]

Chen Fang, Lorenzo Torresani
2012 Lecture Notes in Computer Science  
In this work we introduce novel image metrics that can be used with distance-based classifiers or directly to decide whether two input images belong to the same class.  ...  In a preprocessing stage our approach derives a semantic image graph from the labeled dataset, where the nodes are the labeled images and the edges connect pictures with related labels.  ...  Nowozin and C. Rother for useful discussion on strategies to optimize our SEO energy and to T. Deselaers and V. Ferrari for sharing data. Thanks to A. Bergamo for help with the experiments.  ... 
doi:10.1007/978-3-642-33765-9_29 fatcat:l35duo5d3rgizosczhzyist2qe

Few-Shot Hash Learning for Image Retrieval

Liangke Gui, Yu-Xiong Wang, Martial Hebert
2017 2017 IEEE International Conference on Computer Vision Workshops (ICCVW)  
The need for labeled samples limits their applicability in scenarios in which a user provides at query time a small set of training images defining a customized novel category.  ...  We can then select a task-specific combination of hash codes for a novel category from a few labeled samples.  ...  Given a new target task, e.g., a novel category, at query time with a small set of training images, we could simply use all of these codes as descriptors for image retrieval.  ... 
doi:10.1109/iccvw.2017.148 dblp:conf/iccvw/GuiWH17 fatcat:fsaj4glmgzcgbdzgw75to7vy5i

Visual and semantic similarity in ImageNet

Thomas Deselaers, Vittorio Ferrari
2011 CVPR 2011  
The insights gained from our analysis enable building a novel distance function between images assessing whether they are from the same basic-level category.  ...  Many computer vision approaches take for granted positive answers to questions such as "Are semantic categories visually separable?" and "Is visual similarity correlated to semantic similarity?".  ...  Classemes [39] for object recognition, to combine multiple kernels in [16] , and in [35] where label histograms output by random forests are fed into SVMs.  ... 
doi:10.1109/cvpr.2011.5995474 dblp:conf/cvpr/DeselaersF11 fatcat:km2o6q4iv5fc3oo74y7nv2m2lu

Visual Coding in a Semantic Hierarchy

Yang Yang, Hanwang Zhang, Mingxing Zhang, Fumin Shen, Xuelong Li
2015 Proceedings of the 23rd ACM international conference on Multimedia - MM '15  
In this paper, we present a novel visual coding approach, termed as hierarchical semantic visual coding (HSVC), to effectively encode visual objects (e.g., image and video) in a semantic hierarchy.  ...  In recent years, tremendous research endeavours have been dedicated to seeking effective visual representations for facilitating various multimedia applications, such as visual annotation and retrieval  ...  [11] extended the meaning of group into "class", "tag" and "instance" levels and proposed a multi-layer group sparse coding algorithm for image annotation.  ... 
doi:10.1145/2733373.2806244 dblp:conf/mm/YangZZSL15 fatcat:3xdtzlwcejewhjcrprnv2snrmm

Object Bank: An Object-Level Image Representation for High-Level Visual Recognition

Li-Jia Li, Hao Su, Yongwhan Lim, Li Fei-Fei
2013 International Journal of Computer Vision  
We provide guidelines for effectively applying OB to high level image recognition tasks where it could be easily compressed for efficient computation in practice and is very robust to various classifiers  ...  We introduce the novel concept of object bank (OB), a high-level image representation encoding object appearance and spatial location information in images.  ...  semantic properties whereas classemes focuses more on the compact, efficient properties of an image representation.  ... 
doi:10.1007/s11263-013-0660-x fatcat:e3zgztkvkzeu5hbgpfwnjzumma

Learning Semantic Signatures for 3D Object Retrieval

Boqing Gong, Jianzhuang Liu, Xiaogang Wang, Xiaoou Tang
2013 IEEE transactions on multimedia  
We convert such qualitative descriptions into attribute signature (AS) and reference set signature (RSS), respectively, and use them for 3DOR.  ...  Humans are capable of describing an object using attribute terms like "symmetric" and "flyable", or using its similarities to some known object classes.  ...  In [20] and [31] , reference sets are used as a knowledge transfer model for 2D object recognition. They have been used in face recognition [35] and image retrieval [28] .  ... 
doi:10.1109/tmm.2012.2231059 fatcat:xy2hnq72jvg67jiktirzgoi3sa

Hybrid-Indexing Multi-type Features for Large-Scale Image Search [chapter]

Qingjun Luo, Shiliang Zhang, Tiejun Huang, Wen Gao, Qi Tian
2015 Lecture Notes in Computer Science  
Both of the two indexing models are suited for specific features and are limited to certain scenarios like partial-duplicate search and similar image search, respectively.  ...  Moreover, hybrid-indexing does not degrade the efficiency of classic BoWs based image search. Experiments on several public datasets manifest the effectiveness and efficiency of our proposed method.  ...  Qi Tian by ARO grant W911NF-12-1-0057 and Faculty Research Awards by NEC Laboratories of America. This work was supported in part by National Science Foundation of China (NSFC) 61429201.  ... 
doi:10.1007/978-3-319-16865-4_29 fatcat:vwtkcepgl5hb3aelig6zyloa64
« Previous Showing results 1 — 15 out of 22 results