High-dimensional signature compression for large-scale image classification
The DOI matches its Crossref record; 33 of the 34 references it cites were found in a registry or an index.
Abstract
We address image classification on a large-scale, i.e. when a large number of images and classes are involved. First, we study classification accuracy as a function of the image signature dimensionality and the training set size. We show experimentally that the larger the training set, the higher the impact of the dimensionality on the accuracy. In other words, high-dimensional signatures are important to obtain state-of-the-art results on large datasets. Second, we tackle the problem of data compression on very large signatures (on the order of 105dimensions) using two lossy compression strategies: a dimensionality reduction technique known as the hash kernel and an encoding technique based on product quantizers. We explain how the gain in storage can be traded against a loss in accuracy and/or an increase in CPU cost. We report results on two large databases - ImageNet and a dataset of lM Flickr images - showing that we can reduce the storage of our signatures by a factor 64 to 128 with little loss in accuracy. Integrating the decompression in the classifier learning yields an efficient and scalable training algorithm. On ILSVRC2010 we report a 74.3% accuracy at top-5, which corresponds to a 2.5% absolute improvement with respect to the state-of-the-art. On a subset of 10K classes of ImageNet we report a top-1 accuracy of 16.7%, a relative improvement of 160% with respect to the state-of-the-art.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“since then the best published results are 45.7% and 25.7% with an approach that averages the predictions of two classifiers trained on Fisher Vectors (FVs) computed from two types of densely sampled features.29”
From ImageNet classification with deep convolutional neural networks (2017) · cites this paper as [29]
Citations
Cites
ImageNet: A large-scale hierarchical image database
Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., & Fei-Fei, L. (2009) · 2009 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2009.5206848 · Cited by 65,333 · cited 1×
“org) consists of more than 12M images of 17K concepts [7] and Flickr contains thousands of groups (www.”
BackgroundBeyond Bags of Features: Spatial Pyramid Matching for Recognizing Natural Scene Categories
Lazebnik, S., Schmid, C., & Ponce, J. (2006) · 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Volume 2 (CVPR'06) · doi:10.1109/cvpr.2006.68 · Cited by 7,916 · cited 1×
“[15], one can split an image into several regions, compute one FV per region and concatenate the per-region FVs.”
BackgroundLocality-constrained Linear Coding for image classification
Wang, J., Yang, J., Yu, K., Lv, F., Huang, T., & Gong, Y. (2010) · 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2010.5540018 · Cited by 3,100 · cited 2×
“Maji and Berg [18] and Wang et al. [31] then proposed efficient algorithms to learn IKSVMs.”
ResultProduct Quantization for Nearest Neighbor Search
Jégou, H., Douze, M., & Schmid, C. (2011) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2010.57 · Cited by 2,956 · cited 3×
“[13] proposed product quantizers (PQ) for NN-search of SIFT vectors.”
BackgroundMethodLinear spatial pyramid matching using sparse coding for image classification
Yang, J., Yu, K., Gong, Y., & Huang, T. (2009) · 2009 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2009.5206757 · Cited by 2,828 · cited 1×
“Yang et al. [36], Wang et al. [32] and Boureau et al. [4] showed that replacing the average pooling stage in the BOV computation by a max-pooling yielded excellent results.”
ResultWeiss, Y., Torralba, A. B., & Fergus, R. (2008) · neural information processing systems · Cited by 2,767
Aggregating local descriptors into a compact image representation
Jegou, H., Douze, M., Schmid, C., & Perez, P. (2010) · 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2010.5540039 · Cited by 2,692 · cited 3×
“For instance, [23] requires dense projections and [34, 5, 14] require a global PCA.”
BackgroundMethodImproving the Fisher Kernel for Large-Scale Image Classification
Perronnin, F., Sánchez, J., & Mensink, T. (2010) · Lecture Notes in Computer Science · doi:10.1007/978-3-642-15561-1_11 · Cited by 2,588 · cited 15×
“As shown in [22], square-rooting and L2-normalizing the FV can greatly enhance the classification accuracy.”
BackgroundMethodInfluentialSelf-citationZhang, J., Marszałek, M., Lazebnik, S., & Schmid, C. (2007) · International Journal of Computer Vision · doi:10.1007/s11263-006-9794-4 · Cited by 1,923 · cited 1×
“The winners of the PASCAL VOC 2007 [8] and 2008 [9] competitions used a similar paradigm: many types of low-level local features are extracted (referred to as “channels”), one bag-of-visual-words (BOV) histogram is computed for each channel and non-linear kernel classifiers such as SVMs are used to perform classificat…”
MethodEvaluating Color Descriptors for Object and Scene Recognition
van de Sande, K. E. A., Gevers, T., & Snoek, C. G. M. (2010) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2009.154 · Cited by 1,783 · cited 1×
“The winners of the PASCAL VOC 2007 [8] and 2008 [9] competitions used a similar paradigm: many types of low-level local features are extracted (referred to as “channels”), one bag-of-visual-words (BOV) histogram is computed for each channel and non-linear kernel classifiers such as SVMs are used to perform classificat…”
Method
Research works printed but not linked: 6
- shi, 2009, Hash kernels, AISTATS
- bottou, 0, SGD
- berg, 2010, ILSVRC
- gong, 2009, Image classification using Gaussian mixture and local coordinate coding, PASCAL VOC 2009 Workshop
- everingham, 0, The PASCAL VOC 2008 Results
- everingham, 2007, The PASCAL VOC
Cited by
ImageNet classification with deep convolutional neural networks
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017) · Communications of the ACM · doi:10.1145/3065386 · Cited by 112,155
“7% with an approach that averages the predictions of two classifiers trained on Fisher Vectors (FVs) computed from two types of densely sampled features.(29) We also entered our model in the ILSVRC-2012 competition and report our results in Table 2.”
ImageNet Large Scale Visual Recognition Challenge
Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Z., Karpathy, A., Khosla, A., Bernstein, M., et al. (2015) · International Journal of Computer Vision · doi:10.1007/s11263-015-0816-y · Cited by 39,380
“Im a g eN et L a rg e S ca le V isu a l R eco g n itio n C h a llen g e 1 7 ILSVRC 2010 Codename CLS LOC Insitutions Contributors and references Hminmax 54.4 Massachusetts Institute of Technology Jim Mutch, Sharat Chikkerur, Hristo Paskov, Ruslan Salakhutdinov, Stan Bileschi, Hueihan Jhuang IBM 70.1 IBM research†, Geo…”
BackgroundMethodInfluential(2014) · Cited by 33,419
Improving neural networks by preventing co-adaptation of feature detectors
Hinton, G. E., Srivastava, N., Krizhevsky, A., Sutskever, I., & Salakhutdinov, R. R. (2012) · arXiv · doi:10.48550/arxiv.1207.0580 · Cited by 6,382
“7% [16].”
Deep Convolutional Neural Networks for Image Classification: A Comprehensive Review
Rawat, W., & Wang, Z. (2017) · Neural Computation · doi:10.1162/neco_a_00990 · Cited by 3,503
“Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/neco_a_00990 by guest on 04 June 2021 Lin et al., 2011; Sánchez & Perronnin, 2011), as well as an approximate correlation between classification performance and network depth (see section 5.5.4).”
BackgroundInfluentialBuilding high-level features using large scale unsupervised learning
Le, Q. V. (2013) · 2013 IEEE International Conference on Acoustics, Speech and Signal Processing · doi:10.1109/icassp.2013.6639343 · Cited by 1,695
“We followed the experimental protocols specified by (Deng et al., 2010; Sanchez & Perronnin, 2011), in which, the datasets are randomly split into two halves for training and validation.”
BackgroundMethodImage Classification with the Fisher Vector: Theory and Practice
Sánchez, J., Perronnin, F., Mensink, T., & Verbeek, J. (2013) · International Journal of Computer Vision · doi:10.1007/s11263-013-0636-x · Cited by 1,498
Self-citationDeep Neural Networks for Object Detection
Szegedy, C., Toshev, A., & Erhan, D. (2013) · neural information processing systems · Cited by 1,662
EndoNet: A Deep Architecture for Recognition Tasks on Laparoscopic Videos
Twinanda, A. P., Shehata, S., Mutter, D., Marescaux, J., de Mathelin, M., & Padoy, N. (2017) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2016.2593957 · Cited by 1,096
“In recent studies [18], [19], radio frequency identification (RFID)-tagged surgical tools have been proposed for tool detection and tracking.”
BackgroundWang, J., Zhang, T., Song, J., Sebe, N., & Shen, H. T. (2018) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2017.2699960 · Cited by 971
“to a wide-range of applications such as large scale object retrieval [51], image classification [122] and detection [139].”
BackgroundLabel-Embedding for Image Classification
Akata, Z., Perronnin, F., Harchaoui, Z., & Schmid, C. (2016) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2015.2487986 · Cited by 770
“This includes work on patch encoding (see [40] for a recent comparison), on kernel-based methods [41] with a recent focus on explicit embeddings [42], [43], on dimensionality reduction [41] and on compression [44], [45], [46].”
MethodSelf-citationLabel-Embedding for Attribute-Based Classification
Akata, Z., Perronnin, F., Harchaoui, Z., & Schmid, C. (2013) · 2013 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2013.111 · Cited by 618
“how to represent an image? This includes works on patch encoding (see [4] for a recent comparison), on kernel-based methods [32] with a recent focus on explicit embeddings [20, 35], on dimensionality reduction [32] and on compression [13, 30, 36].”
BackgroundSelf-citationBuilding high-level features using large scale unsupervised learning
Ranzato, M., Monga, R., Devin, M., Chen, K., Corrado, G. S., Dean, J., Le, Q. V., & Ng, A. Y. (2012) · International Conference on Machine Learning · Cited by 823
Chen, D., Cao, X., Wen, F., & Sun, J. (2013) · 2013 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2013.389 · Cited by 615
“[31] reported on the significance of high-dimensional features in large-scale image classification.”
BackgroundLearning Locally-Adaptive Decision Functions for Person Verification
Li, Z., Chang, S., Liang, F., Huang, T. S., Cao, L., & Smith, J. R. (2013) · 2013 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2013.463 · Cited by 466
“Any Mahalanobis distance between x and y can be regarded as the Euclidean distance on the space transformed by L , namely, d ( x, y ) = ( x − y ) t M ( x − y ) = ( x − y ) L t L ( x − y ) = (cid:4) x (cid:2) − y (cid:2) (cid:4) 22 , where M = L t L , x (cid:2) = Lx and y (cid:2) = Ly .”
BackgroundFisher Vector Faces in the Wild
Simonyan, K., Parkhi, O., Vedaldi, A., & Zisserman, A. (2013) · Procedings of the British Machine Vision Conference 2013 · doi:10.5244/c.27.8 · Cited by 445
“Dense feature extraction is an essential component of many state-of-the-art image classification methods [18, 23, 27].”
BackgroundDistance-Based Image Classification: Generalizing to New Classes at Near-Zero Cost
Mensink, T., Verbeek, J., Perronnin, F., & Csurka, G. (2013) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2013.83 · Cited by 431
“The current state-of-the-art [1], [2] uses efficient linear SVM classifiers trained in a one-vs-rest manner in combination with high-dimensional bag-of-words [18], [19] or Fisher vector representations [16].”
BackgroundMethodResultInfluentialSelf-citationComputer vision algorithms and hardware implementations: A survey
Feng, X., Jiang, Y., Yang, X., Du, M., & Li, X. (2019) · Integration · doi:10.1016/j.vlsi.2019.07.005 · Cited by 349
Additive Quantization for Extreme Vector Compression
Babenko, A., & Lempitsky, V. (2014) · 2014 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2014.124 · Cited by 278
“Firstly, image compression can be applied to training images in order to facilitate learning on very large image collections when the whole training data needs to be loaded into the CPU [19, 22].”
BackgroundMethodMetric Learning for Large Scale Image Classification: Generalizing to New Classes at Near-Zero Cost
Mensink, T., Verbeek, J., Perronnin, F., & Csurka, G. (2012) · Lecture Notes in Computer Science · doi:10.1007/978-3-642-33709-3_35 · Cited by 281
Self-citationFace Anti-Spoofing using Speeded-Up Robust Features and Fisher Vector Encoding
Boulkenafet, Z., Komulainen, J., & Hadid, A. (2016) · IEEE Signal Processing Letters · doi:10.1109/lsp.2016.2630740 · Cited by 275
“Extracting dense features has shown to be an essential component in many computer vision applications [24], [25].”
BackgroundEffects of Image Degradation and Degradation Removal to CNN-Based Image Classification
Pei, Y., Huang, Y., Zou, Q., Zhang, X., & Wang, S. (2021) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2019.2950923 · Cited by 253
“While image classification [1], [2] has been studied in different applications for a long time, its performance is substantially improved in recent years by using supervised deep learning, e.”
BackgroundMethodEfficient Feature Extraction, Encoding, and Classification for Action Recognition
Kantorov, V., & Laptev, I. (2014) · 2014 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2014.332 · Cited by 239
“We next investigate Fisher vector (FV) encoding [23] and show improved action recognition while using fast linear classifiers.”
MethodBuilding high-level features using large scale unsupervised learning
Le, Q. V., Ranzato, M., Monga, R., Devin, M., Chen, K., Corrado, G. S., Dean, J., & Ng, A. Y. (2011) · arXiv (Cornell University) · doi:10.48550/arxiv.1112.6209 · Cited by 163
“We followed the experimental protocols specified by (Deng et al., 2010; Sanchez & Perronnin, 2011), in which, the datasets are randomly split into two halves for training and validation.”
BackgroundMethodDeep CNN-based visual defect detection: Survey of current literature
Jha, S. B., & Babiceanu, R. F. (2023) · Computers in Industry · doi:10.1016/j.compind.2023.103911 · Cited by 224
Multi-Instance Deep Learning: Discover Discriminative Local Anatomies for Bodypart Recognition
Yan, Z., Zhan, Y., Peng, Z., Liao, S., Shinagawa, Y., Zhang, S., Metaxas, D. N., & Zhou, X. S. (2016) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2016.2524985 · Cited by 222
“The framework of global representations followed by classical classifiers has been widely used in scene recognition [22] and image classification [23], [24].”
MethodHedging your bets: Optimizing accuracy-specificity trade-offs in large scale visual recognition
Deng, J., Krause, J., Berg, A. C., & Fei-Fei, L. (2012) · 2012 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2012.6248086 · Cited by 166
“Our work is also inspired by an emerging trend in computer vision studying large scale visual recognition [19, 27, 13, 12, 31, 7, 26, 24].”
BackgroundLearning Binary Codes for High-Dimensional Data Using Bilinear Projections
Gong, Y., Kumar, S., Rowley, H. A., & Lazebnik, S. (2013) · 2013 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2013.69 · Cited by 164
“To achieve high retrieval and classification accuracy on such datasets, it is necessary to use extremely high-dimensional descriptors such as Fisher Vectors (FV) [23, 24, 28], Vectors of Locally Aggregated Descriptors (VLAD) [14], or Locality Constrained Linear Codes (LLC) [32].”
BackgroundMethodInfluentialScaling the Scattering Transform: Deep Hybrid Networks
Oyallon, E., Belilovsky, E., & Zagoruyko, S. (2017) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2017.599 · Cited by 202
“On the other hand prior to 2012 local encoding meth-ods led to state of the art performance on large scale visual recognition tasks [30].”
InfluentialInteractive plant identification based on social image data
Joly, A., Goëau, H., Bonnet, P., Bakić, V., Barbe, J., Selmi, S., Yahiaoui, I., Carré, J., Mouysset, E., Molino, J.-F., et al. (2014) · Ecological Informatics · doi:10.1016/j.ecoinf.2013.07.006 · Cited by 158
“Sánchez and Perronnin (2011) reported a top-1 accuracy of 54.3% on thousand categories (74.3% top-5 accuracy).”
BackgroundInfluentialGood Practice in Large-Scale Learning for Image Classification
Akata, Z., Perronnin, F., Harchaoui, Z., & Schmid, C. (2014) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2013.146 · Cited by 158
“Large-scale image classification has recently received significant interest from the computer vision and machine learning communities [4], [20], [22], [48], [62], [66], [72], [82], [89].”
BackgroundSelf-citationCirculant Binary EmbeddingPreprint
Yu, F., Kumar, S., Gong, Y., & Chang, S. (2014) · arXiv (Cornell University) · doi:10.48550/arxiv.1405.3162 · Cited by 153
“The advantage is to save on storage allowing even large scale datasets to fit in memory (Li et al., 2011; Sánchez & Perronnin, 2011).”
BackgroundMethodInfluentialIncremental Learning of Random Forests for Large-Scale Image Classification
Ristin, M., Guillaumin, M., Gall, J., & Van Gool, L. (2016) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2015.2459678 · Cited by 154
Deep Fisher Networks for Large-Scale Image Classification
Simonyan, K., Vedaldi, A., & Zisserman, A. (2013) · Oxford University Research Archive (ORA) (University of Oxford) · Cited by 207
Fisher vectors meet Neural Networks: A hybrid classification architecture
Perronnin, F., & Larlus, D. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298998 · Cited by 132
“In 2010-2011, the leading approaches to ILSVRC [43, 31] were based on the Fisher Vector [38, 39] (FV) or the closely related Super Vector (SV) [57].”
BackgroundMethodInfluentialSelf-citationLinear Regression-Based Efficient SVM Learning for Large-Scale Classification
Wu, J., & Yang, H. (2015) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2014.2382123 · Cited by 134
“Efforts have been made in extracting higher dimensional features (see in [9], [22]), and we expect PmSVM to achieve higher accuracies using such more complex feature sets.”
BackgroundMethodSegmentation Driven Object Detection with Fisher Vectors
Cinbis, R. G., Verbeek, J., & Schmid, C. (2013) · 2013 IEEE International Conference on Computer Vision · doi:10.1109/iccv.2013.369 · Cited by 117
Image classification using Deep learning
krishna, M. M., Neelima, M., Harshali, M., & Rao, M. V. G. (2018) · International Journal of Engineering & Technology · doi:10.14419/ijet.v7i2.7.10892 · Cited by 140
Towards good practice in large-scale learning for image classification
Perronnin, F., Akata, Z., Harchaoui, Z., & Schmid, C. (2012) · 2012 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2012.6248090 · Cited by 101
“Large-scale image classification has recently received significant interest [11, 19, 28, 29].”
BackgroundMethodInfluentialSelf-citationMulti-loss Regularized Deep Neural Network
Xu, C., Lu, C., Liang, X., Gao, J., Zheng, W., Wang, T., & Yan, S. (2016) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2015.2477937 · Cited by 93
DeepChart: Combining deep convolutional networks and deep belief networks in chart classification
Tang, B., Liu, X., Lei, J., Song, M., Tao, D., Sun, S., & Dong, F. (2015) · Signal Processing · doi:10.1016/j.sigpro.2015.09.027 · Cited by 87
Sparse projections for high-dimensional binary codes
Xia, Y., He, K., Kohli, P., & Sun, J. Z. (2015) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2015.7298954 · Cited by 87
“In computer vision, methods for generating binary representations have been successfully used for problems like image retrieval [31, 21, 11], image classifi-cation [28, 10, 32], and descriptor matching [17].”
BackgroundMethodLarge-scale image classification with trace-norm regularization
Harchaoui, Z., Douze, M., Paulin, M., Dudík, M., & Malick, J. (2012) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2012.6248078 · Cited by 113
“As an example, the winning approaches of 2010 edition of the ImageNet Large Scale Visual Recognition Challenge (ILSVRC)(1) were built upon one-vs-rest classifiers [19, 26].”
BackgroundMethodInfluentialDetecting Avocados to Zucchinis: What Have We Done, and Where Are We Going?
Russakovsky, O., Deng, J., Huang, Z., Berg, A. C., & Fei-Fei, L. (2013) · 2013 IEEE International Conference on Computer Vision · doi:10.1109/iccv.2013.258 · Cited by 82
Incremental Learning of NCM Forests for Large-Scale Image Classification
Ristin, M., Guillaumin, M., Gall, J., & Van Gool, L. (2014) · 2014 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2014.467 · Cited by 82
“To address the problem of large data, there has been a wide use of online learning methods, such as stochastic gradient descent [13, 20].”
BackgroundSparse kernel approximations for efficient classification and detection
Vedaldi, A., & Zisserman, A. (2012) · 2012 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2012.6247943 · Cited by 82
“Each image is represented by the Fisher encoding xi obtained as described in [19]: first, dense SIFT features are extracted every three pixels (using the vl_phow function of [26]) and compressed from dimension 128 to 80 by using PCA; then a Gaussian mixture model with 256 components is fitted to sample PCA-SIFT descri…”
BackgroundMethodInfluentialTransformation Pursuit for Image Classification
Paulin, M., Revaud, J., Harchaoui, Z., Perronnin, F., & Schmid, C. (2014) · 2014 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2014.466 · Cited by 80
“We then use PQ compression [13] which was shown to yield an almost-negligible loss of performance in large-scale classification [24].”
MethodSelf-citationVLAD3: Encoding Dynamics of Deep Features for Action Recognition
Li, Y., Li, W., Mahadevan, V., & Vasconcelos, N. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.215 · Cited by 83
“Over the last decade, the dominant representation for both problems was the bag-of-words model, combining histograms of descriptors such as HoG or HoF and pooling over space-time volumes [24, 32].”
BackgroundSegmentation-free word spotting with exemplar SVMs
Almazán, J., Gordo, A., Fornés, A., & Valveny, E. (2014) · Pattern Recognition · doi:10.1016/j.patcog.2014.06.005 · Cited by 96
Scattering Networks for Hybrid Representation Learning
Oyallon, E., Zagoruyko, S., Huang, G., Komodakis, N., Lacoste-Julien, S., Blaschko, M. B., & Belilovsky, E. (2018) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2018.2855738 · Cited by 112
Probabilistic Label Trees for Efficient Large Scale Image Classification
Liu, B., Sadeghi, F., Tappen, M. F., Shamir, O., & Liu, C. (2013) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2013.114 · Cited by 78
Investigation of transfer learning for image classification and impact on training sample size
Zhu, W., Braun, B., Chiang, L. H., & Romagnoli, J. A. (2021) · Chemometrics and Intelligent Laboratory Systems · doi:10.1016/j.chemolab.2021.104269 · Cited by 83
Training highly multiclass classifiers
Gupta, M. R., Bengio, S., & Weston, J. (2014) · Journal of Machine Learning Research · doi:10.5555/2627435.2638582 · Cited by 101
Jointly Learning Visually Correlated Dictionaries for Large-Scale Visual Recognition Applications
Zhou, N., & Jianping Fan, B. (2013) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2013.189 · Cited by 67
“Index Terms—Joint dictionary learning, common visual atoms, category-specific visual atoms, visual tree, large-scale visual recognition Ç”
BackgroundLocal similarity-aware deep feature embedding
Huang, C., Loy, C. C., & Tang, X. (2016) · neural information processing systems · Cited by 67
Efficient Action Localization with Approximately Normalized Fisher Vectors
Oneaţă, D., Verbeek, J. J., & Schmid, C. (2014) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2014.326 · Cited by 76
“[14, 24].”
Rao, J., Qiao, Y., Ren, F., Wang, J., & Du, Q. (2017) · Sensors · doi:10.3390/s17091951 · Cited by 72
“Since Krizhevsky et al. [55] won the classification task of ILSVRC 2012 with the lowest top-5 error rate (15.3%, much lower than that of the traditional machine learning method SIFT+FVs [56], which was 26.2”
ResultInfluentialA Testbed for Cross-Dataset Analysis
Tommasi, T., & Tuytelaars, T. (2015) · Lecture notes in computer science · doi:10.1007/978-3-319-16199-0_2 · Cited by 86
Common Subspace for Model and Similarity: Phrase Learning for Caption Generation from Images
Ushiku, Y., Yamaguchi, M., Mukuta, Y., & Harada, T. (2015) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.306 · Cited by 76
“Such linear weight vectors are also employed in combination with high-dimensional image feature [37], and in the full connected layers in CNN [17].”
BackgroundMethodMulti-class Multi-annotator Active Learning with Robust Gaussian Process for Visual Recognition
Long, C., & Hua, G. (2015) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.325 · Cited by 74
“Most of the current recognition systems are based on supervised learning with large quantity of labeled training data [15, 29].”
BackgroundA Deep Convolution Neural Network Model for Vehicle Recognition and Face Recognition
Luo, X., Shen, R., Hu, J., Deng, J., Hu, L., & Guan, Q. (2017) · Procedia Computer Science · doi:10.1016/j.procs.2017.03.153 · Cited by 70
Efficient Exemplar Word Spotting
Almazán, J., Gordo, A., Fornés, A., & Valveny, E. (2012) · British Machine Vision Conference (BMVC) · doi:10.5244/c.26.67 · Cited by 63
“Similarly to [20], we decode the features and learn our classifier on the decoded data.”
MethodJoint Hierarchical Category Structure Learning and Large-Scale Image Classification
Qu, Y., Lin, L., Shen, F., Lu, C., Wu, Y., Xie, Y., & Tao, D. (2016) · IEEE Transactions on Image Processing · doi:10.1109/tip.2016.2615423 · Cited by 73
“We next compare our method with seven representative image classification methods: HOG + LBP + sparse coding [34], SIFT + Fisher vector [47], Fisher vector [44], one-vs.”
MethodUsing Multi-level Convolutional Neural Network for Classification of Lung Nodules on CT images
Lyu, J., & Ling, S. H. (2018) · Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) · doi:10.1109/embc.2018.8512376 · Cited by 63
“Sai Ho Ling is with the School of Biomedical Engineering, University of Technology Sydney, Ultimo, NSW, Australia (e-mail: Steve.Ling@uts.edu.au). coding [9] and scale-invariant feature transform plus Fisher Vectors [10] (SIFT+FVs).”
BackgroundActive Visual Recognition with Expertise Estimation in Crowdsourcing
Long, C., Hua, G., & Kapoor, A. (2013) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2013.373 · Cited by 57
“Introduction As research on visual recognition evolving more towards an experimental science, partly due to the success of the introduction of machine learning approach to computer vision [21, 13, 14], collecting labeled visual datasets at large scale from crowd-sourcing tools such as Amazon Mechanical Turk has become…”
BackgroundUnclonable Photonic Crystal Hydrogels with Controllable Encoding Capacity for Anticounterfeiting
Wu, J., Li, J., Liu, X., Gong, L., Chen, J., Tang, Z., Lin, W., Mu, Y., Lin, X., Hong, W., et al. (2021) · ACS Applied Materials & Interfaces · doi:10.1021/acsami.1c20905 · Cited by 56
Power mean SVM for large scale visual classification
Wu, J. (2012) · 2012 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2012.6247946 · Cited by 54
“In the future, we may explore an approach as that in [21]: compress the dataset and handle the compressed data on the fly.”
Deep learning for object recognition: A comprehensive review of models and algorithms
Tsirtsakis, P., Zacharis, G., Maraslidis, G. S., & Fragulis, G. F. (2025) · International Journal of Cognitive Computing in Engineering · doi:10.1016/j.ijcce.2025.01.004 · Cited by 102
Long, C., Hua, G., & Kapoor, A. (2015) · International Journal of Computer Vision · doi:10.1007/s11263-015-0834-9 · Cited by 50
“…theintro-duction of the machine learning approach to computer vision (Burl et al. 1995; Burl and Perona 1998; Fergus et al. 2005; Liu et al. 2008; Sanchez and Perronnin 2011; Krizhevsky et al. 2012; Lin et al. 2011), collecting labeled visual datasets at large scale from crowd-sourcing tools…”
BackgroundMeta-class features for large-scale object categorization on a budget
Bergamo, A., & Torresani, L. (2012) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2012.6248040 · Cited by 55
Collaborative Active Visual Recognition from Crowds: A Distributed Ensemble Approach
Hua, G., Long, C., Yang, M., & Gao, Y. (2017) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2017.2682082 · Cited by 45
“Nevertheless, stateof-the-art recognition systems are all based on supervised learningwith large amount of labeled training data [22], [34].”
BackgroundJoint Multi-View Hashing for Large-Scale Near-Duplicate Video Retrieval
Nie, X., Jing, W., Cui, C., Zhang, C., Zhu, L., & Yin, Y. (2019) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2019.2913383 · Cited by 52
“Existing hashing techniques can be coarsely divided into two main categories: data-independent [6], [10], [11] and data-dependent hashing [12], [13], [14], [15].”
BackgroundStructure-Aware Deep Learning for Product Image Classification
Chen, Z., Ai, S., & Jia, C. (2019) · ACM Transactions on Multimedia Computing Communications and Applications · doi:10.1145/3231742 · Cited by 46
“However, the best performing method still returned a top-5 error rate of more than 25% [33].”
MethodCompact Representation for Image Classification: To Choose or to Compress?
Zhang, Y., Wu, J., & Cai, J. (2014) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2014.121 · Cited by 50
“We use the Fisher Vector to represent all images, following the setup in [22].”
BackgroundMethodInfluentialBi-directional Heterogeneous Graph Hashing towards Efficient Outfit Recommendation
Guan, W., Song, X., Zhang, H., Liu, M., Yeh, C., & Chang, X. (2022) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3503161.3548020 · Cited by 57
“Learning to hash, which aims to learn a compact binary hash code for the given instance, has attracted a large amount of research interest [24, 28], due to its prominent advantages in saving the time and storage costs.”
BackgroundStructural color‐based physical unclonable function
Lin, X., Shi, D., Yi, G., & Yu, D. (2024) · Responsive materials · doi:10.1002/rpm.20230031 · Cited by 55
Collaborative Active Learning of a Kernel Machine Ensemble for Recognition
Hua, G., Long, C., Yang, M., & Gao, Y. (2013) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2013.153 · Cited by 46
“Supervised discriminative learning has been one of the main methodologies for advancing research on visual recognition [20, 13].”
MethodFast Orthogonal Projection Based on Kronecker Product
Zhang, X., Yu, F. X., Guo, R., Kumar, S., Wang, S., & Chang, S.-F. (2015) · 2015 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.335 · Cited by 43
“In practice, to achieve high recall in retrieval tasks, it is often desirable to use long codes with a large k such that k = O(d) [22, 8, 29].”
Large-Scale Aerial Image Categorization Using a Multitask Topological Codebook
Zhang, L., Wang, M., Hong, R., Yin, B., & Li, X. (2015) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2015.2408592 · Cited by 43
“The fish vector (FV) [55] extends the BoW by counting the zero-order statistics and by encoding the firstand second-order statistics of the distribution of descriptors assigned to each visual word.”
BackgroundVISOR: Towards On-the-Fly Large-Scale Object Category Retrieval
Chatfield, K., & Zisserman, A. (2013) · Lecture notes in computer science · doi:10.1007/978-3-642-37444-9_34 · Cited by 49
“This method has been shown to produce excellent performance both for the image classification task [20] and for large-scale retrieval [20, 7].”
BackgroundMethodInfluentialOn-the-fly learning for visual search of large-scale image and video datasets
Chatfield, K., Arandjelović, R., Parkhi, O., & Zisserman, A. (2015) · International Journal of Multimedia Information Retrieval · doi:10.1007/s13735-015-0077-0 · Cited by 48
“We explore the use of product quantization (PQ), which has been widely used as a compression method for image features [26,44] and works by splitting the original feature into Q-dimensional sub-blocks, each of which is encoded using a separate vocabulary of cluster centres pre-learned from a training set.”
MethodHow to use Bag-of-Words model better for image classification
Wang, C., & Huang, K. (2014) · Image and Vision Computing · doi:10.1016/j.imavis.2014.10.013 · Cited by 44
Object detection and recognition via clustered features
Woźniak, M., & Połap, D. (2018) · Neurocomputing · doi:10.1016/j.neucom.2018.09.003 · Cited by 39
The Unbearable Shallow Understanding of Deep Learning
Plebe, A., & Grasso, G. M. (2019) · Minds and Machines · doi:10.1007/s11023-019-09512-8 · Cited by 49
Joint Hierarchical Category Structure Learning and Large-Scale Image Classification
Qu, Y., Lin, L., Shen, F., Lu, C., Wu, Y., Xie, Y., & Tao, D. (2017) · UTS ePRESS (University of Technology Sydney) · Cited by 39
Value-laden disciplinary shifts in machine learning
Dotan, R., & Milli, S. (2020) · Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency · doi:10.1145/3351095.3373157 · Cited by 37
DEFEATnet—A Deep Conventional Image Representation for Image Classification
Gao, S., Duan, L., & Tsang, I. W.-H. (2015) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2015.2389413 · Cited by 40
“It is worth noting that on the very challenging ImageNet classification task (ILSVRC-2012), convolutional neural net (CNN)-based DNN outperforms conventional image representation methods (SIFT + FVs [12]) by about 8% in terms of recognition accuracy [13], and achieves the best performance on this task.”
MethodExploiting artificial intelligence for digitally enriched museum visits
Ioannakis, G., Bampis, L., & Koutsoudis, A. (2019) · Journal of Cultural Heritage · doi:10.1016/j.culher.2019.07.019 · Cited by 46
Classemes and Other Classifier-Based Features for Efficient Object Categorization
Bergamo, A., & Torresani, L. (2014) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2014.2313111 · Cited by 39
“Over the last few years this topic has received a growing amount of attention in the vision community [1], [2].”
Liu, Q., Liu, G., Li, L., Yuan, X., Wang, M., & Liu, W. (2017) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2017.2696053 · Cited by 32
Zhang, X., Zhang, Y., Qian, B., Liu, X., Li, X., Wang, X., Yin, C., Lv, X., Song, L., & Wang, L. (2019) · Lecture notes in computer science · doi:10.1007/978-3-030-17938-0_19 · Cited by 36
Steno, P., Alsadoon, A., Prasad, P. W. C., Al-Dala’in, T., & Alsadoon, O. H. (2020) · The Journal of Supercomputing · doi:10.1007/s11227-020-03418-4 · Cited by 31
“On ILSVRC-2010 competition, the model achieved results superior to the state-of-the-art solution of Sánchez and Perronnin [14].”
ResultJoint multi-feature spatial context for scene recognition in the semantic manifold
Song, X., Jiang, S., & Herranz, L. (2015) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2015.7298736 · Cited by 29
“…Scene Recognition in the Semantic Manifold Xinhang Song, Shuqiang Jiang, Luis Herranz Key Laboratory of Intelligent Information Processing of Chinese Academy of Sciences (CAS) Institute of Computer Technology, CAS, Beijing, 100190, China {xinhang.song, shuqiang.jiang, luis.herranz}@vipl.ict.ac.cn”
BackgroundUnified Simultaneous Clustering and Feature Selection for Unlabeled and Labeled Data
Han, D., & Kim, J. (2018) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2018.2818444 · Cited by 31
“Feature compression methods, such as [1]–[3], compress the original feature values to 0 or 1 by quantization.”
MethodMaking Online Sketching Hashing Even Faster
Chen, X., Yang, H., Zhao, S., Lyu, M. R.-T., & King, I. (2019) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2019.2934687 · Cited by 28
“HASHING on features is an efficient tool to conduct approximate nearest neighbor searches for many machine learning applications such as large scale object retrieval [24], image classification [44], fast object detection [16], and image matching [15], etc.”
Suzuki, G., Takahashi, S., Ogawa, T., & Haseyama, M. (2019) · IEEE Access · doi:10.1109/access.2019.2946378 · Cited by 32
“network (CNN) [32], [33], which is one of the deep learning techniques, have achieved more accurate recognition than that achieved by traditional machine learning-based recognition methods [25], [34].”
BackgroundTo what extent do DNN-based image classification models make unreliable inferences?
Tian, Y., Ma, S., Wen, M., Liu, Y., Cheung, S.-C., & Zhang, X. (2021) · Empirical Software Engineering · doi:10.1007/s10664-021-09985-1 · Cited by 44
“While these models outperform classic algorithms, such as SIFT+FV (Sanchez and Perronnin 2011) and Sparse Coding (Lin et al. 2011), in terms of classification accuracy (Krizhevsky et al. 2012), which is the proportion of the inputs in test set whose inference result is the same as the ground truth, recent studies have…”
MethodEfficient image annotation for automatic sentence generation
Ushiku, Y., Harada, T., & Kuniyoshi, Y. (2012) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/2393347.2393424 · Cited by 31
“In [28, 14], highdimensional features in [25, 32] are used for learning 1000 classes from over a million images.”
BackgroundMethodFast action retrieval from videos via feature disaggregation
Qin, J., Liu, L., Yu, M., Wang, Y., & Shao, L. (2016) · Computer Vision and Image Understanding · doi:10.1016/j.cviu.2016.09.009 · Cited by 29
Hybrid multi-layer deep CNN/aggregator feature for image classification
Kulkarni, P., Zepeda, J., Jurie, F., Pérez, P., & Chevallier, L. (2015) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp.2015.7178196 · Cited by 28
“The performance gap of DCNNs relative to the second entry in that year’s competition (and relative to SIFT-based Fisher aggregation schemes [19]) is in excess of 10 percentage points in absolute improvement of top-5 error rate.”
Background
Metadata
- Authors
- Sanchez, J., & Perronnin, F.
- Affiliations
- Xerox (France) (FR)
- Venue
- CVPR 2011
- Publisher
- IEEE
- Published
- 2011-06
- Pages
- 1665-1672
- ISSN
- 1063-6919 (ISSN-L)
- OpenAlex
- W2053229256
- Semantic Scholar
- eefcc7bcc05436dac9881acb4ff4e4a0b730e175
- Cited by
- 272
- References
- 34
- Citations by year
- 2026: 3 · 2025: 5 · 2024: 7 · 2023: 6 · 2022: 11 · 2021: 19 · 2020: 10 · 2019: 23 · 2018: 17 · 2017: 23 · 2016: 21 · 2015: 46 · 2014: 30 · 2013: 30 · 2012: 18
- FWCI
- 19.28
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings (publishedVersion)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Image and Video Retrieval Techniques
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Image Retrieval and Classification Techniques
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Keywords
- image classification, large-scale image classification, signature compression, lossy compression, dimensionality reduction, training data size, storage efficiency, scalable training, ImageNet, Flickr images
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.