Fast image search for learned metrics

By Prateek Jain, Brian Kulis, Kristen Grauman. 2008 IEEE Conference on Computer Vision and Pattern Recognition, 2008

319 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/11e7d074e4deebfed771d0aa226f4160f4c5c82b

How researchers see this paper

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

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Learning query-dependent prefilters for scalable image retrieval (2009, background): “[7] extend this representation to compute a learned distance metric (hq−hi)A(hq−hi), which can be optimized with the same efficiency, and thus their problem statement is very close to ours.”
Max-margin additive classifiers for detection (2009, methodology): “For image classification problems, recent results have made approximate nearest neighbor techniques in high dimensional feature spaces, that require no training but may learn parameters for hashing, efficient enough to be used [38, 19, 35, 22] .”
Metric Learning: A Survey (2013, background): “The work in [38, 43] explored how to compute these functions when running metric learning in kernel space, for particular kernels, resulting in a method that alternates between updating hash functions and updating the learned metrics.”
Linear spatial pyramid matching using sparse coding for image classification (2009, methodology): “However, their performances are still behind the state-of-the-art achieved by [12, 1, 9] on public benchmarks.”
Local features are not lonely – Laplacian sparse coding for image classification (2010, methodology): “In our work, we adopt the widely used SIFT[11] feature due to its excellent performance in image recognition[2, 20, 1, 6].”
Linear cross-modal hashing for efficient multimedia search (2013, background): “Copyrights for components of this work owned by others than ACM must be honored.”
Learning to Hash for Indexing Big Data—A Survey (2015, background): “Other interesting learning ideas include twostep learning methods which treat hash bit learning and hash function learning separately [85][86].”
Inter-media hashing for large-scale retrieval from heterogeneous data sources (2013, methodology): “In [17], LSH is combined with a learnt Mahalanobis metric which is learnt from a labeled image dataset to reflect semantic information.”
Data fusion through cross-modality metric learning using similarity-sensitive hashing (2010, methodology): “A similar approach is used in [14] in a visual content-based search application.”
Deep learning of binary hash codes for fast image retrieval (2015, background): “Several hashing algorithms [6, 29, 18, 20, 28, 10] have been proposed to approximately identify data relevant to the query.”
Learning to Hash with Binary Reconstructive Embeddings (2009, background): “Locality-sensitive hashing [1, 2] is the most popular metho d, and extensions have been explored for accommodating distances such as lp norms [10], learned metrics [11], and image kernels [12].”
Kernelized locality-sensitive hashing for scalable image search (2009, background): “Previously, hash functions have been designed for cases where the similarity function " sim " refers to an p norm, Mahalanobis metric, or inner product [6] [16] [3].”
Scalable Nearest Neighbor Algorithms for High Dimensional Data (2014, methodology): “computed using various learning techniques: parameter sensitive hashing [5], spectral hashing [32], randomized LSH hashing from learned metrics [33], kernelized LSH [34], learnt binary embeddings [35], shift-invariant kernel hashing [36], semi-supervised hashing [37], optimized kernel hashing [38] a”
Learning mid-level features for recognition (2010, background): “[9] Fast image search for learned metrics 61.”
Geometric min-Hashing: Finding a (thick) needle in a haystack (2009, methodology): “[12] introduced a method for efficient extension of Locally Sensitive Hashing scheme [11] for Mahalanobis distance.”