Revisiting Additive Quantization

Martínez, Julieta; Clement, Joris; Hoos, Holger H.; Little, James J.

doi:10.1007/978-3-319-46475-6_9

Cited by 41 publications

(38 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Thanks to both high compression quality and computational efficiency PQ-based methods are currently the top choice for compact representations of large datasets. PQ gave rise to active research on high-dimensional vectors compression in computer vision and machine learning community [11,12,13,14,15,16,17,18,19]. IVFADC [1] is one of the first retrieval systems capable of dealing with billion-scale datasets efficiently.…”

Section: Related Workmentioning

confidence: 99%

Revisiting the Inverted Indices for Billion-Scale Approximate Nearest Neighbors

Baranchuk

Babenko

Malkov

2018

Lecture Notes in Computer Science

View full text Add to dashboard Cite

This work addresses the problem of billion-scale nearest neighbor search. The state-of-the-art retrieval systems for billion-scale databases are currently based on the inverted multi-index, the recently proposed generalization of the inverted index structure. The multi-index provides a very fine-grained partition of the feature space that allows extracting concise and accurate short-lists of candidates for the search queries. In this paper, we argue that the potential of the simple inverted index was not fully exploited in previous works and advocate its usage both for the highly-entangled deep descriptors and relatively disentangled SIFT descriptors. We introduce a new retrieval system that is based on the inverted index and outperforms the multi-index by a large margin for the same memory consumption and construction complexity. For example, our system achieves the state-of-the-art recall rates several times faster on the dataset of one billion deep descriptors compared to the efficient implementation of the inverted multi-index from the FAISS library. * https://github.com/dbaranchuk/ivf-hnsw

show abstract

Section: Related Workmentioning

confidence: 99%

Revisiting the Inverted Indices for Billion-Scale Approximate Nearest Neighbors

Baranchuk

Babenko

Malkov

2018

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

“…To the best of our knowledge, sparse DSQ is the first attempt to explore supervised sparse multicodebook quantization for semantic similarity search. Nevertheless, we compare our technique with two unsupervised sparse quantization techniques, SCQ [47] and SLSQ [29] applied to the deep features of the fc7 layer of the deep model in [10].…”

Section: Resultsmentioning

confidence: 99%

“…Since the former criterion imposes a harder sparsity constraint on the codebooks, we would naturally expect to achieve lower search accuracy but better query time. We compare against SCQ1 and SCQ2 from [47] and SLSQ1 and SLSQ2 from [29]. Figure 3 shows the performance of different techniques against three different datasets.…”

Section: Resultsmentioning

confidence: 99%

Deep Spherical Quantization for Image Search

Eghbali

Tahvildari

2019

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

View full text Add to dashboard Cite

Hashing methods, which encode high-dimensional images with compact discrete codes, have been widely applied to enhance large-scale image retrieval. In this paper, we put forward Deep Spherical Quantization (DSQ), a novel method to make deep convolutional neural networks generate supervised and compact binary codes for efficient image search. Our approach simultaneously learns a mapping that transforms the input images into a low-dimensional discriminative space, and quantizes the transformed data points using multi-codebook quantization. To eliminate the negative effect of norm variance on codebook learning, we force the network to L 2 normalize the extracted features and then quantize the resulting vectors using a new supervised quantization technique specifically designed for points lying on a unit hypersphere. Furthermore, we introduce an easyto-implement extension of our quantization technique that enforces sparsity on the codebooks. Extensive experiments demonstrate that DSQ and its sparse variant can generate semantically separable compact binary codes outperforming many state-of-the-art image retrieval methods on three benchmarks.

show abstract

“…Method VOC2007 Caltech-101 ImageNet PQ [24] 0.4965 0.3089 0.1650 CKM [38] 0.4995 0.3179 0.1737 LSQ [37] 0.4993 0.3372 0.1882 DSH-64 [33] 0.4914 0.2852 0.1665 SUBIC 2-layer 0.5600 0.3923 0.2543 SUBIC 3-layer 0.5588 0.4033 0.2810 Table 3: Cross-domain category retrieval. Performance (mAP) using 64-bit encoders across three different datasets using VGG-128 as base feature extractor.…”

Section: Methodsmentioning

confidence: 99%

SuBiC: A Supervised, Structured Binary Code for Image Search

Jain

Zepeda

Pérez

et al. 2017

2017 IEEE International Conference on Computer Vision (ICCV)

View full text Add to dashboard Cite

We introduce SUBIC, a supervised structured binary code (concatenation of one-hot blocks). It is produced by a novel supervised deep convolutional network and is well adapted to efficient visual search, including category retrieval for which it outperforms the state-of-the-art supervised deep binary hashing techniques. AbstractFor large-scale visual search, highly compressed yet meaningful representations of images are essential. Structured vector quantizers based on product quantization and its variants are usually employed to achieve such compression while minimizing the loss of accuracy. Yet, unlike binary hashing schemes, these unsupervised methods have not yet benefited from the supervision, end-to-end learning and novel architectures ushered in by the deep learning revolution. We hence propose herein a novel method to make deep convolutional neural networks produce supervised, compact, structured binary codes for visual search. Our method makes use of a novel block-softmax nonlinearity and of batch-based entropy losses that together induce structure in the learned encodings. We show that our method outperforms state-of-the-art compact representations based on deep hashing or structured quantization in single and cross-domain category retrieval, instance retrieval and classification. We make our code and models publicly available online.

show abstract

Revisiting Additive Quantization

Cited by 41 publications

References 24 publications

Revisiting the Inverted Indices for Billion-Scale Approximate Nearest Neighbors

Revisiting the Inverted Indices for Billion-Scale Approximate Nearest Neighbors

Deep Spherical Quantization for Image Search

SuBiC: A Supervised, Structured Binary Code for Image Search

Contact Info

Product

Resources

About