Searching with quantization: approximate nearest neighbor search using short codes and distance estimators

Hervé Jégou 1 Matthijs Douze 1 Cordelia Schmid 1
1 LEAR - Learning and recognition in vision
Inria Grenoble - Rhône-Alpes, LJK - Laboratoire Jean Kuntzmann, INPG - Institut National Polytechnique de Grenoble
Abstract : We propose an approximate nearest neighbor search method based on quantization. It uses, in particular, product quantizer to produce short codes and corresponding distance estimators approximating the Euclidean distance between the orginal vectors. The method is advantageously used in an asymmetric manner, by computing the distance between a vector and code, unlike competing techniques such as spectral hashing that only compare codes. Our approach approximates the Euclidean distance based on memory efficient codes and, thus, permits efficient nearest neighbor search. Experiments performed on SIFT and GIST image descriptors show excellent search accuracy. The method is shown to outperform two state-of-the-art approaches of the literature. Timings measured when searching a vector set of 2 billion vectors are shown to be excellent given the high accuracy of the method.
Document type :
Reports
Complete list of metadatas

Cited literature [28 references]  Display  Hide  Download


https://hal.inria.fr/inria-00410767
Contributor : Hervé Jégou <>
Submitted on : Monday, August 24, 2009 - 12:57:37 PM
Last modification on : Tuesday, June 18, 2019 - 3:38:03 PM
Long-term archiving on : Monday, October 15, 2012 - 4:21:05 PM

Files

RR-7020.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : inria-00410767, version 1

Collections

Citation

Hervé Jégou, Matthijs Douze, Cordelia Schmid. Searching with quantization: approximate nearest neighbor search using short codes and distance estimators. [Research Report] RR-7020, INRIA. 2009, pp.25. ⟨inria-00410767⟩

Share

Metrics

Record views

1443

Files downloads

1506