Bundling features for large scale partial-duplicate web image search

By Zhong Wu, Qifa Ke, Michael Isard. 2009

451 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

Aggregating Local Image Descriptors into Compact Codes (2012, background): “Typical applications include finding locations [1] and particular objects [2], as well as detecting partial image duplicates on the web [3] and deformed copies [4].”
Local features are not lonely – Laplacian sparse coding for image classification (2010, background): “However, recent work shows that the relationship among the features are important for image representation, such as the geometric relationship [23].”
Scalar quantization for large scale image search (2012, methodology): “The geometric clues, such as feature position, scale, and orientation, are also stored in inverted file list for verification of geometric consistency [3][4][15][16][17].”
Sketch-Based Image Retrieval: Benchmark and Bag-of-Features Descriptors (2011, methodology): “While the use of local descriptors is a common approach in example-based image retrieval [4], [5], [6], [7], SBIR systems up to now still employ global descriptors and thus inherit their drawbacks, mainly being not invariant to affine transformations.”
To Aggregate or Not to aggregate: Selective Match Kernels for Image Search (2013, unknown): “This is done by multiple assignment [23, 12], or by exploiting a more precise representation of the individual local descriptors, such as binary codes in the so-called Hamming Embedding (HE) method [12], or by integrating some information about the neighborhood of the descriptor [31].”
Image retrieval with geometry-preserving visual phrases (2011, background): “Cooccurrences in the entire images [18] do not capture spatial information between the words, while co-occurrences in local neighborhoods [21, 25, 20] capture neighboring information, but ignore long-range interactions.”
Predicting Good Features for Image Geo-Localization Using Per-Bundle VLAD (2015, methodology): “From an input query image with unknown geo-location (a), MSER regions and SIFT keypoints forms a bundled feature [40], and consequently represented by PBVLAD (b).”
Sketch-based manga retrieval using manga109 dataset (2015, methodology): “This kind of task has been tackled in spatial veriﬁcation-based reranking methods [25], [26], [27], [28], [29], [30], [31].”
SIFT Meets CNN: A Decade Survey of Instance Retrieval (2016, methodology): “Visual phrases can also be constructed from adjacent image patches [103], random spatial partitioning [106], and localized stable regions [29] such as MSER [28].”