Ensemble of exemplar-SVMs for object detection and beyond
Open accessThe DOI matches its Crossref record; 23 of the 25 references it cites were found in a registry or an index.
Abstract
This paper proposes a conceptually simple but surprisingly powerful method which combines the effectiveness of a discriminative object detector with the explicit correspondence offered by a nearest-neighbor approach. The method is based on training a separate linear SVM classifier for every exemplar in the training set. Each of these Exemplar-SVMs is thus defined by a single positive instance and millions of negatives. While each detector is quite specific to its exemplar, we empirically observe that an ensemble of such Exemplar-SVMs offers surprisingly good generalization. Our performance on the PASCAL VOC detection task is on par with the much more complex latent part-based model of Felzenszwalb et al., at only a modest computational cost increase. But the central benefit of our approach is that it creates an explicit association between each detection and a single training exemplar. Because most detections show good alignment to their associated exemplar, it is possible to transfer any available exemplar meta-data (segmentation, geometric structure, 3D model, etc.) directly onto the detections, which can then be used as part of overall scene understanding.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Figures


![Figure 3 Comparison. Given a bicycle training sample from PASCAL (represented with a HOG weight vector w ), we show the top 6 matches from the PASCAL test-set using three methods. Row 1: naive nearest neighbor (using raw normalized HOG). Row 2: Trained Exemplar-SVM (notice how w focuses on bike-specific edges). Row 3: Learned distance function - an Exemplar-SVM but trained in the 'distance-to-exemplar' vector space, with the exemplar being placed at the origin (loosely corresponding to [10, 14]).](https://figures.cochat.ai/paper-figures/1cc4b956c3534a219577db00fbbb8da4/ensemble-of-exemplar-svms-for-object-detection-and-beyond-figure-3.png)







Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
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
“In the case of sliding-window object detectors, such as the deformable parts model (DPM) [12], this imbalance may be as extreme as 100,000 background examples to every one object.The recent trend towards object-proposal-based detectors [15,32] mitigates this issue to an extent, but the imbalance ratio may still be high (e.g., 70:1).”
From Training Region-Based Object Detectors with Online Hard Example Mining (2016) · cites this paper as [32]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
LIBSVM: A library for support vector machines
Chang, C.-C., & Lin, C.-J. (2011) · ACM Transactions on Intelligent Systems and Technology · doi:10.1145/1961189.1961199 · Cited by 40,918 · cited 1×
“Using the procedure above, we train an ensemble of Exemplar-SVM, one for each positive instance in the training set.”
MethodHistograms of Oriented Gradients for Human Detection
Dalal, N., & Triggs, B. (2005) · 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) · doi:10.1109/cvpr.2005.177 · Cited by 33,584 · cited 2×
“Yet today, a number of methods can do just that with reasonable accuracy.”
BackgroundThe pascal visual object classes (VOC) challenge
Everingham, M., Van Gool, L., Williams, C. K. I., Winn, J., & Zisserman, A. (2010) · International Journal of Computer Vision · doi:10.1007/s11263-009-0275-4 · Cited by 19,599 · cited 2×
“For our experiments, we use a single source of exemplars: the PASCAL VOC 2007 dataset [8] – a popular dataset used to benchmark object detection algorithms.”
BackgroundMethodLibsvm : A library for support vector machines
Ferrari, V. (2008) · Medical Entomology and Zoology · Cited by 10,122
Object Detection with Discriminatively Trained Part-Based Models
Felzenszwalb, P. F., Girshick, R. B., McAllester, D., & Ramanan, D. (2010) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2009.167 · Cited by 9,853 · cited 4×
“For this, we present results on a set of tasks including segmentation, qualitative-geometry estimation, 3D model transfer, and related object priming.”
BackgroundMethodInfluentialProbabilistic Outputs for Support vector Machines and Comparisons to Regularized Likelihood Methods
Platt, J. C. (1999) · Cited by 6,384
Shape matching and object recognition using shape contexts
Belongie, S., Malik, J., & Puzicha, J. (2002) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/34.993558 · Cited by 6,306 · cited 1×
“Because most detections show good alignment to their associated exemplar, it is possible to transfer any available exemplar meta-data (segmentation, geometric structure, 3D model, etc.) directly onto the detections, which can then be used as part of overall scene understanding.”
BackgroundEstimating the Support of a High-Dimensional Distribution
Schölkopf, B., Platt, J. C., Shawe-Taylor, J., Smola, A. J., & Williamson, R. C. (2001) · Neural Computation · doi:10.1162/089976601750264965 · Cited by 6,168
SUN database: Large-scale scene recognition from abbey to zoo
Xiao, J., Hays, J., Ehinger, K. A., Oliva, A., & Torralba, A. (2010) · 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2010.5539970 · Cited by 3,283 · cited 1×
“Because most detections show good alignment to their associated exemplar, it is possible to transfer any available exemplar meta-data (segmentation, geometric structure, 3D model, etc.) directly onto the detections, which can then be used as part of overall scene understanding.”
Background80 Million Tiny Images: A Large Data Set for Nonparametric Object and Scene Recognition
Torralba, A., Fergus, R., & Freeman, W. T. (2008) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2008.128 · Cited by 1,921
Research works printed but not linked: 1
- A. Frome and J. Malik. Image retrieval and recognition using local distance functions. NIPS, 2006. 2, 3, 4
Cited by
Items 101–200 of 873
Domain Generalization and Adaptation Using Low Rank Exemplar SVMs
Li, W., Xu, Z., Xu, D., Dai, D., & Van Gool, L. (2018) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2017.2704624 · Cited by 125
“In contrast to [28], we introduce a nuclear norm based regularizer on the prediction matrix in order to exploit the low-rank structure from latent domains for domain generalization.”
MethodA Survey of Deep Learning-Based Object Detection Methods and Datasets for Overhead Imagery
Kang, J., Tariq, S., Oh, H., & Woo, S. S. (2022) · IEEE Access · doi:10.1109/access.2022.3149052 · Cited by 131
“Their proposed training process enables the model to be robust in various domains by learning domain-invariant features.”
BackgroundAnalysis of the Hands in Egocentric Vision: A Survey
Bandini, A., & Zariffa, J. (2020) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2020.2986648 · Cited by 140
“Serra et al. [36] classified the binary segmentation masks into multiple hand configurations by using an ensemble of exemplar-SVMs [133].”
MethodUnsupervised Feature Learning for Writer Identification and Writer Retrieval
Christlein, V., Gropp, M., Fiel, S., & Maier, A. (2017) · IAPR International Conference on Document Analysis and Recognition (ICDAR) · doi:10.1109/icdar.2017.165 · Cited by 137
“This method was originally proposed for object detection [35], where an ensemble of E-SVMs is used for each object class.”
MethodWhat makes instance discrimination good for transfer learning?
Zhao, N., Wu, Z., Lau, R. W. H., & Lin, S. (2020) · arXiv · doi:10.48550/arxiv.2006.06606 · Cited by 115
“Unlike the original exemplar SVM which represents positives non-parametrically and negatives parametrically, our pretraining scheme models all instances in an non-parametric fashion in a spirit similar to instance discrimination [35].”
BackgroundMethodInfluentialObject Tracking with Multi-View Support Vector Machines
Zhang, S., Yu, X., Sui, Y., Zhao, S., & Zhang, L. (2015) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2015.2390044 · Cited by 120
Writer Identification Using GMM Supervectors and Exemplar-SVMs
Christlein, V., Bernecker, D., Hönig, F., Maier, A., & Angelopoulou, E. (2016) · Pattern Recognition · doi:10.1016/j.patcog.2016.10.005 · Cited by 131
“The authors of [20] showed that an ensemble of such Exemplar-SVMs generalizes well, although each single Exemplar-SVM has a very strict decision boundary.”
BackgroundMethodInfluentialSelf Paced Deep Learning for Weakly Supervised Object Detection
Sangineto, E., Nabi, M., Culibrk, D., & Sebe, N. (2019) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2018.2804907 · Cited by 114
“[7] use Exemplar SVMs (ESVMs) [26] to train a classifier from a single positive sample.”
MethodWatch and learn: Semi-supervised learning of object detectors from videos
Misra, I., Shrivastava, A., & Hebert, M. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298982 · Cited by 104
“We start tracking from the confident ESVM detections to label new examples.”
BackgroundMethodInfluentialCastaldo, F., Zamir, A., Angst, R., Palmieri, F., & Savarese, S. (2015) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw.2015.137 · Cited by 123
Hara, K., Sun, J., Moore, R. J., Jacobs, D. W., & Froehlich, J. E. (2014) · ACM Symposium on User Interface Software and Technology (UIST) · doi:10.1145/2642918.2647403 · Cited by 132
“Second, we post-process the resulting bounding boxes using non-maximum suppression [37] and 3D-point cloud data to eliminate”
MethodTowards Computational Baby Learning: A Weakly-Supervised Approach for Object Detection
Liang, X., Liu, S., Wei, Y., Liu, L., Lin, L., & Yan, S. (2015) · 2015 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.120 · Cited by 88
“Inspired by [25], we train a separate linear SVM classifier for each positive instance, and each SVM classifier is highly tuned according to the exemplar’s appearance.”
Enriching Visual Knowledge Bases via Object Discovery and Segmentation
Chen, X., Shrivastava, A., & Gupta, A. (2014) · 2014 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2014.261 · Cited by 105
“Specifically, we train an exemplar detector [10, 24] (eLDA in our case) based on Color-HOG (CHOG) features [31].”
MethodSelf-citationFigure-ground segmentation by transferring window masks
Kuettel, D., & Ferrari, V. (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.2012.6247721 · Cited by 115
“Afterwards, the energy minimization stage combines the local 1 evidence of all windows into a coherent global segmentation of the whole test image.”
BackgroundExpanded Parts Model for Human Attribute and Action Recognition in Still Images
Sharma, G., Jurie, F., & Schmid, 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.2013.90 · Cited by 106
“The recently proposed Exemplar SVM (ESVM) [21] learns discriminative templates for each object instance of the training set independently and then combines their calibrated outputs on test images as a post-processing step.”
MethodBeyond Hard Negative Mining: Efficient Detector Learning via Block-Circulant Decomposition
Henriques, J. F., Carreira, J., Caseiro, R., & Batista, J. (2013) · 2013 IEEE International Conference on Computer Vision · doi:10.1109/iccv.2013.343 · Cited by 107
“Hard negative mining is considered expensive, and this has become less tolerable as the community has strived to scale up to a large number of object models [22].”
BackgroundAction Recognition with Exemplar Based 2.5D Graph Matching
Yao, B., & Fei-Fei, L. (2012) · Lecture notes in computer science · doi:10.1007/978-3-642-33765-9_13 · Cited by 102
“While the majority of work in computer vision are model based, exemplar based methods have also been applied in object recognition [25–27] and video classification [28].”
BackgroundLearning and Calibrating Per-Location Classifiers for Visual Place Recognition
Gronát, P., Obozinski, G., Šivic, J., & Pajdla, T. (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.2013.122 · Cited by 111
Deep Exemplar 2D-3D Detection by Adapting from Real to Rendered Views
Massa, F., Russell, B. C., & Aubry, M. (2016) · 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.2016.648 · Cited by 106
“Examples include alignment of IKEA furniture models to images [36], exemplar-based object detection [38] by matching discriminative elements [3, 9], and using hand-crafted features for retrieving CAD models for depth prediction [48] and compositing from multiple models [27].”
MethodStocco, A., Yandrapally, R., & Mesbah, A. (2018) · ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE) · doi:10.1145/3236024.3236063 · Cited by 100
“In brief, (1) the matches that do not fall in the region where the key-points have been found are discarded, and (2) a non-maxima suppression operation (NMS) is also applied [39] (basically NMS assumes highly overlapping detections as belonging to the same object [9]).”
BackgroundEfficient Boosted Exemplar-Based Face Detection
Li, H., Lin, Z., Brandt, J., Shen, X., & Hua, G. (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.2014.238 · Cited by 101
“In related research areas, the work presented in [17, 23, 13] are conceptually related to our approach as they combined the boosting framework with discriminatively trained exemplar classifiers.”
MethodME R-CNN: Multi-Expert R-CNN for Object Detection
Lee, H., Eum, S., & Kwon, H. (2019) · IEEE Transactions on Image Processing · doi:10.1109/tip.2019.2938879 · Cited by 106
“One of the conventional ways to generate RoIs is to use multi-scale sliding windows [2, 8, 12, 15, 22, 27, 35] which can be considered as a ‘dense’ search.”
BackgroundMethodNonparametric Part Transfer for Fine-Grained Recognition
Goering, C., Rodner, E., Freytag, A., & Denzler, 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.2014.319 · Cited by 114
“We want to point out that our part detection technique is indeed related to the idea of Exemplar-SVM [23].”
BackgroundMethodInfluentialEX2: Exploration with Exemplar Models for Deep Reinforcement Learning
Fu, J., Co-Reyes, J. D., & Levine, S. (2017) · arXiv · doi:10.48550/arxiv.1703.01260 · Cited by 95
“We borrow the term “exemplar model” from Malisiewicz et al. (2011), which coined the term “ex-emplar SVM” to refer to a particular linear model trained to classify each instance against all others.”
MethodTilted Empirical Risk MinimizationPreprint
Li, T., Beirami, A., Sanjabi, M., & Smith, V. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2007.01162 · Cited by 101
“We compare TERM with several baselines which re-weight samples during training, including focal loss [37], HardMine [38], and LearnReweight [50].”
BackgroundInfluentialArticulated Pose Estimation Using Discriminative Armlet Classifiers
Gkioxari, G., Arbelaez, P., Bourdev, L., & Malik, J. (2013) · 2013 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2013.429 · Cited by 91
“The classifiers used were nearest neighbors and we have evidence that discriminative classifiers such as support vector machines (SVMs) typically outdo nearest neighbor approaches [15].”
MethodSubcategory-Aware Object Classification
Dong, J., Xia, W., Chen, Q., Feng, J., Huang, Z., & Yan, S. (2013) · 2013 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2013.112 · Cited by 92
“However, each of these methods has its own limitations and ignores other more general intra-class variations such as appearance and shape variance [27, 18].”
BackgroundMethodInfluentialEfficient facial expression recognition using histogram of oriented gradients in wavelet domain
Nigam, S., Singh, R. K., & Misra, A. K. (2018) · Multimedia Tools and Applications · doi:10.1007/s11042-018-6040-3 · Cited by 95
“SVM multiclass classifier provides better competitiveness and scalability on large data [6, 42].”
BackgroundContext-Aware Convolutional Neural Network for Object Detection in VHR Remote Sensing Imagery
Gong, Y., Xiao, Z., Tan, X., Sui, H., Xu, C., Duan, H., & Li, D. (2019) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2019.2930246 · Cited by 99
Domain-Adaptive Discriminative One-Shot Learning of Gestures
Pfister, T., Charles, J., & Zisserman, A. (2014) · Lecture notes in computer science · doi:10.1007/978-3-319-10599-4_52 · Cited by 107
“Other costs (e.g . spatial alignments, SVM training and testing, subtitle preprocessing etc.) are negligible in comparison (a few seconds per gesture/video).”
BackgroundMethodInfluentialMultiple queries for large scale specific object retrieval
Arandjelović, R., & Zisserman, A. (2012) · British Machine Vision Conference (BMVC) · doi:10.5244/c.26.92 · Cited by 92
The State of the Art: Object Retrieval in Paintings using Discriminative Regions
Crowley, E. J., & Zisserman, A. (2014) · British Machine Vision Conference (BMVC) · doi:10.5244/c.28.38 · Cited by 107
“…paintings, in particular for the case of a specific building, rather than a class of objects: Shrivastava et al. [32] use an Exemplar SVM [26] for retrieval, and Aubry et al. [6] improve on this method by employing mid-level discriminative patches (MLDPs) [7, 23, 27, 33] to allow for more variation.”
Li, Y., Liu, L., Shen, C., & van den Hengel, A. (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.2015.7298699 · Cited by 110
“Note that only training images are used to discover patterns.”
BackgroundIntelligent Visual Media Processing: When Graphics Meets Vision
Cheng, M., Hou, Q., Zhang, S., & Rosin, P. L. (2017) · Journal of Computer Science and Technology · doi:10.1007/s11390-017-1681-7 · Cited by 91
Prasad, K. N. R. S. V., Dsouza, K. B., & Bhargava, V. K. (2020) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2020.2987990 · Cited by 89
“(ii) wt ∈ [24 , 32 , 40 , 48 , 56 , 64 , 72 , 80 , 160] , corresponding to signals of time-span [0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1, 2]ms respectively, (iii) pix ∈ [2 , 4 , 6 , 8 , 16] , (iv) the block size b = 2 × pix (each block is therefore a grid of 4 cells), the block stride b stride = 0 .”
MethodInfluentialAnnotating metabolite mass spectra with domain-inspired chemical formula transformers
Goldman, S., Wohlwend, J., Stražar, M., Haroush, G., Xavier, R. J., & Coley, C. W. (2023) · Nature Machine Intelligence · doi:10.1038/s42256-023-00708-3 · Cited by 87
Deep learning based single sample face recognition: a survey
Liu, F., Chen, D., Wang, F., Li, Z., & Xu, F. (2022) · Artificial Intelligence Review · doi:10.1007/s10462-022-10240-2 · Cited by 95
CliqueCNN: Deep Unsupervised Exemplar Learning
Bautista, M. Á., Sanakoyeu, A., Sutter, E., & Ommer, B. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1608.08792 · Cited by 81
“However, large intra-class variability has recently spurred exemplar methods [15, 11], which split this problem into simpler sub-tasks.”
MethodInfluentialLearning Data Manipulation for Augmentation and WeightingPreprint
Hu, Z., Tan, B., Salakhutdinov, R., Mitchell, T. M., & Xing, E. P. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1910.12795 · Cited by 115
“, augmenting by flipping an image or replacing a word with synonyms, and weighting with inverse class frequency or loss values [10, 32].”
BackgroundCollaborative Deep Reinforcement Learning for Joint Object Search
Kong, X., Xin, B., Wang, Y., & Hua, G. (2017) · 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.2017.748 · Cited by 94
Pano2Vid: Automatic Cinematography for Watching 360 $$^{\circ }$$ Videos
Su, Y.-C., Jayaraman, D., & Grauman, K. L. (2017) · Lecture notes in computer science · doi:10.1007/978-3-319-54190-7_10 · Cited by 82
“SVM [43], each algorithm-generated video is assigned a ranking based on its distance from the decision boundary (i.”
Object detection through search with a foveated visual system
Akbaş, E., & Eckstein, M. P. (2017) · PLoS Computational Biology · doi:10.1371/journal.pcbi.1005743 · Cited by 91
“We use 2 templates per class while DCC trains more than 15 templates per class within an exemplar-SVM [29]-like framework.”
BackgroundMethodInfluentialSegmentation-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
“Keywords: word spotting, segmentation-free, unsupervised learning, reranking, query expansion, compression”
BackgroundOptasia: A Relational Platform for Efficient Large-Scale Video Analytics
Lu, Y., Chowdhery, A., & Kandula, S. (2016) · Proceedings of the Seventh ACM Symposium on Cloud Computing · doi:10.1145/2987550.2987564 · Cited by 80
“Hence, we use heuristics based on the fixed viewpoint of surveillance cameras (e.g. typical size of objects of interest, lane annotations etc.) and an exemplar SVM [52] to further break the segments.”
BackgroundMethodInteraction part mining: A mid-level approach for fine-grained action recognition
Zhou, Y., Ni, B., Hong, R., Wang, M., & Tian, Q. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298953 · Cited by 77
“Secondly, recent image mid-level patch mining approaches [9, 28, 3, 15] show success in image recognition tasks.”
BackgroundNonparametric image parsing using adaptive neighbor sets
Eigen, D., & Fergus, R. (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.2012.6248004 · Cited by 87
Occlusion and Motion Reasoning for Long-Term Tracking
Yang, H., Alahari, K., & Schmid, C. (2014) · Lecture notes in computer science · doi:10.1007/978-3-319-10599-4_12 · Cited by 84
“This idea of maintaining multiple detectors for an object category is similar in spirit to exemplar SVMs [31].”
Depth edge detection by image-based smoothing and morphological operations
Hasan, S. M. A., & Ko, K. H. (2016) · Journal of Computational Design and Engineering · doi:10.1016/j.jcde.2016.02.002 · Cited by 77
Object Tracking Methods:A Review
Soleimanitaleb, Z., Keyvanrad, M. A., & Jafari, A. (2019) · International Conference on Computer and Knowledge Engineering · doi:10.1109/iccke48569.2019.8964761 · Cited by 91
HWNet v2: an efficient word image representation for handwritten documents
Krishnan, P., & Jawahar, C. V. (2019) · International Journal on Document Analysis and Recognition (IJDAR) · doi:10.1007/s10032-019-00336-x · Cited by 89
“[3] uses an exemplar-svm [44] for representing a query and performs the initial scoring of candidate words.”
MethodNovel Views of Objects from a Single Image
Rematas, K., Nguyen, C. H., Ritschel, T., Fritz, M., & Tuytelaars, T. (2016) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2016.2601093 · Cited by 76
Motion Part Regularization: Improving action recognition via trajectory group selection
Ni, B., Moulin, P., Yang, X., & Yan, S. (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.2015.7298993 · Cited by 86
“Inspired by the recent success of exemplar-SVM [11], for each database motion part, we learn a detector, by taking itself as the only positive training sample and randomly sampling some negative samples from the database.”
MethodRan, Q., Wang, Q., Zhao, B., Wu, Y., Pu, S., & Li, Z. (2021) · IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · doi:10.1109/jstars.2021.3079968 · Cited by 75
“Traditional object detection methods (HOG [1], SVM [2], DPM [3], etc.) generally include a region proposal, feature extraction, and classification, resulting in a low detection efficiency and poor accuracy due to complex procedures, a large number of redundant windows and the poor robustness of…”
MethodRahman, M. M., Tan, Y., Xue, J., & Lü, K. (2019) · IEEE Transactions on Image Processing · doi:10.1109/tip.2019.2955239 · Cited by 78
“To overcome this difficulty, Sliding Shapes [88] exploited depth information in a data-driven fashion and trained exemplar SVM [89] on 3D voxel grids encoded with hand-crafted geometric features by rendering a collection of 3D CAD models from different viewpoints into synthetic depth maps.”
MethodInfluentialLatent-Class Hough Forests for 6 DoF Object Pose Estimation
Tejani, A., Kouskouridas, R., Doumanoglou, A., Tang, D., & Kim, T. (2017) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2017.2665623 · Cited by 81
Mid-level Deep Pattern Mining∗
Li, Y., Liu, L., Shen, C., & van den Hengel, A. (2016) · Cited by 72
Dual Contradistinctive Generative Autoencoder
Parmar, G., Li, D., Lee, K., & Tu, Z. (2021) · 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr46437.2021.00088 · Cited by 68
“Inspired by the literature on instance-level classification [50], approximating likelihood by classification [60], and contrastive learning [20, 65, 22], we propose to model instance-level fidelity by contrastive loss (commonly referred to as InfoNCE loss) [62].”
CMW-Net: Learning a Class-Aware Sample Weighting Mapping for Robust Deep Learning
Shu, J., Yuan, X., Meng, D., & Xu, Z. (2023) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2023.3271451 · Cited by 93
“Typical methods include Boosting and AdaBoost [27], [28], hard negative mining [29] and focal loss [30].”
MethodSemantic alignment of LiDAR data at city scale
Yu, F., Xiao, J., & Funkhouser, T. A. (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.2015.7298781 · Cited by 80
“The following section provides further descriptions of these ideas and details of their implementations.”
BackgroundLearning Collections of Part Models for Object Recognition
Endres, I., Shih, K. J., Jiaa, J., & Hoiem, D. (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.2013.126 · Cited by 76
“Other competitive object detection methods [5] [6] [10] [15] [21] that are supervised by bounding boxes differ primarily in how they automatically organize and align examples.”
Radar-PointGNN: Graph Based Object Recognition for Unstructured Radar Point-cloud Data
Svenningsson, P., Fioranelli, F., & Yarovoy, A. G. (2021) · IEEE Radar Conference (RadarConf) · doi:10.1109/radarconf2147009.2021.9455172 · Cited by 84
“The proposals are thresholded based on a predicted clutter-score 1 and spatially overlapping proposals are suppressed based on non-maximum suppression [11].”
BackgroundAREA: Adaptive Reweighting via Effective Area for Long-Tailed Classification
Chen, X., Zhou, Y., Wu, D., Yang, C., Li, B., Hu, Q., & Wang, W. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.01766 · Cited by 101
“Therefore, it is a serious challenge to train a classifier with a severely unbalanced dataset [35, 19, 36, 1].”
BackgroundAre all training examples equally valuablePreprint
Lapedriza, À., Pirsiavash, H., Bylinskii, Z., & Torralba, A. B. (2013) · arXiv (Cornell University) · doi:10.13016/m2jp7z-giib · Cited by 84
Video Representation Learning Using Discriminative Pooling
Wang, J., Cherian, A., Porikli, F., & Gould, S. J. (2018) · 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.2018.00126 · Cited by 73
Does Negative Sampling Matter? a Review With Insights Into its Theory and Applications
Yang, Z., Ming, D., Huang, T., Cen, Y., Song, J., Xu, B., Dong, Y., & Tang, J. (2024) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2024.3371473 · Cited by 90
“For example, DNS [19] randomly samples a subset as the pool of negatives for probability calculations.”
BackgroundWeakly Supervised Object Localization Using Size Estimates
Shi, M., & Ferrari, V. (2016) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-46454-1_7 · Cited by 67
“Object class detection has been intensively studied during recent years [1,2,3,4,5,6,7,8,9].”
BackgroundMulticlass semantic video segmentation with object-level active inference
Liu, B., & He, X. (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.2015.7299057 · Cited by 72
Local Metric Learning for Exemplar-Based Object Detection
You, X., Li, Q., Tao, D., Ou, W., & Gong, M. (2014) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2014.2306031 · Cited by 70
Drones and deep learning produce accurate and efficient monitoring of large-scale seabird colonies
Hayes, M. C., Gray, P. C., Harris, G., Sedgwick, W. C., Crawford, V. D., Chazal, N., Crofts, S., & Johnston, D. W. (2021) · Ornithological Applications · doi:10.1093/ornithapp/duab022 · Cited by 72
Crowley, E. J., & Zisserman, A. (2016) · Lecture notes in computer science · doi:10.1007/978-3-319-46604-0_50 · Cited by 80
“There is however, work on the specific problem of learning from natural images and retrieving paintings: Shrivastava et al. [36] use an Exemplar-SVM [28] to retrieve paintings of specific buildings.”
MethodSemantic Concept Co-Occurrence Patterns for Image Annotation and Retrieval
Feng, L., & Bhanu, B. (2015) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2015.2469281 · Cited by 67
“The discriminative model is created by an ensemble of instance-SVMs for each concept where the idea is similar to [49].”
MethodMulti-component Models for Object Detection
Gu, C., Arbeláez, P., Lin, Y., Yu, K., & Malik, J. (2012) · Lecture notes in computer science · doi:10.1007/978-3-642-33765-9_32 · Cited by 73
“model also outperforms [17] that trains each component model using only one positive instance.”
BackgroundMethodResultInfluentialDetection, Location and Grasping Objects Using a Stereo Sensor on UAV in Outdoor Environments
Soria, P. R., Arrue, B. C., & Ollero, A. (2017) · Sensors · doi:10.3390/s17010103 · Cited by 66
“Some authors use machine learning algorithms as neural networks [5], or classifiers such as Support Vector Machine (SVM) [6] using a bag-of-words model.”
Exemplar based Deep Discriminative and Shareable Feature Learning for scene image classification
Zuo, Z. C., Wang, G., Shuai, B., Zhao, L., & Yang, Q. (2015) · Pattern Recognition · doi:10.1016/j.patcog.2015.02.003 · Cited by 65
“Comparing to current existing feature learning methods, they can generally get better local descriptors, and extra information (e.g. discriminative information) can be better expressed by manually inserting prior knowledge.”
BackgroundMulti-view Domain Generalization for Visual Recognition
Niu, L., Li, W., & Xu, D. (2015) · 2015 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.477 · Cited by 66
“We observe that ESVM is generally better than SVM, which shows that it is beneficial to fuse multiple exemplar classifiers for enhancing the domain generalization capacity.”
BackgroundMethodInfluentialSingh, K. K., Xiao, F., & Lee, Y. J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.386 · Cited by 63
Few-Shot Object Recognition from Machine-Labeled Web Images
Xu, Z., Zhu, L., & Yang, Y. (2017) · 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.2017.569 · Cited by 73
“Note that on 5-shot and 10-shot tasks, our model achieves better performance than the E-SVMs with larger margin.”
MethodInfluentialAn active search strategy for efficient object class detection
Gonzalez-Garcia, A., Vezhnevets, A., & Ferrari, V. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298921 · Cited by 59
“Hence our approach requires the same annotation as standard object detectors [8, 10, 19, 22, 26, 35, 49, 53].”
BackgroundMethodEfficient 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
“Last, we propose to compress the HOG descriptors by means of PQ.”
BackgroundMethodInfluentialReal-time scalable 6DOF pose estimation for textureless objects
Cao, Z., Sheikh, Y., & Banerjee, N. K. (2016) · 2016 IEEE International Conference on Robotics and Automation (ICRA) · doi:10.1109/icra.2016.7487396 · Cited by 60
“based method [18] to train mid-level visual elements for estimation of identies and poses of chairs in images.”
BackgroundMethodMulti-Correlation Filters With Triangle-Structure Constraints for Object Tracking
Ruan, W., Chen, J., Wu, Y., Wang, J., Liang, C., Hu, R., & Jiang, J. (2019) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2018.2872897 · Cited by 66
“Hence, we introduce an effective part selection scheme by cropping out the local parts with high energy and integrity via superpixel division and exemplar-SVM [39], which is generally applicable to partbased tracking models.”
MethodMultiscale Object Detection from Drone Imagery Using Ensemble Transfer Learning
Walambe, R. A., Marathe, A., & Kotecha, K. V. (2021) · Drones · doi:10.3390/drones5030066 · Cited by 73
“The HOG detector has long been part of many OD algorithms [20–22] and several CV applications.”
MethodTowards Discovery and Attribution of Open-world GAN Generated Images
Girish, S., Suri, S., Rambhatla, S. S., & Shrivastava, A. (2021) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.01383 · Cited by 94
Learning coarse-to-fine sparselets for efficient object detection and scene classification
Cheng, G., Han, J., Guo, L., & Liu, T. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298721 · Cited by 60
“In recent years, since the seminal work of [4, 5] and with the success of deformable part models (DPMs) [6, 7], part model-based methods have demonstrated impressive performance for object detection [6-13] and scene classification [14-21].”
BackgroundJoint Learning of Object and Action Detectors
Kalogeiton, V. S., Weinzaepfel, P., Ferrari, V., & Schmid, C. (2017) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2017.219 · Cited by 64
“Over the past few years, the methods range from low-level features [16, 20, 26, 32, 42, 45, 46], structured models that mine mid-level elements [19, 25] to parts [7, 28, 33] and attributes [22].”
BackgroundMangewa, L. J., Ndakidemi, P. A., & Munishi, L. K. (2019) · Sustainability · doi:10.3390/su11216116 · Cited by 62
Dense Semantic Correspondence Where Every Pixel is a Classifier
Bristow, H., Valmadre, J., & Lucey, S. (2015) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.458 · Cited by 64
Unifying Heterogeneous Classifiers With Distillation
Vongkulbhisal, J., Vinayavekhin, P., & Visentini-Scarzanella, M. (2019) · 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.2019.00329 · Cited by 72
“, multi-class classification [15], object detection [34, 27], tracking [1], etc.”
BackgroundAnalysing Domain Shift Factors between Videos and Images for Object Detection
Kalogeiton, V. S., Ferrari, V., & Schmid, C. (2016) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2016.2551239 · Cited by 60
“Traditionally, training an object detector requires gathering a large, diverse set of still images, in which objects are manually annotated by a bounding-box [8], [15], [17], [28], [42], [44], [46].”
Deep Reinforcement Learning: A State-of-the-Art Walkthrough
Lazaridis, A., Fachantidis, A., & Vlahavas, I. P. (2020) · Journal of Artificial Intelligence Research · doi:10.1613/jair.1.12412 · Cited by 78
“, 2017), the core of which is based in exemplar models (Malisiewicz et al., 2011).”
BackgroundMethodAttention-Based Multi-Level Feature Fusion for Object Detection in Remote Sensing Images
Dong, X., Qin, Y., Gao, Y., Fu, R., Liu, S., & Ye, Y. (2022) · Remote Sensing · doi:10.3390/rs14153735 · Cited by 59
“Finally, Support Vector Machines (SVM) [30] is applied to these features to recognize the object categories within each proposal.”
MethodRegion-based colour modelling for joint crop and maize tassel segmentation
Lü, H., Cao, Z., Xiao, Y., Li, Y., & Zhu, Y. (2016) · Biosystems Engineering · doi:10.1016/j.biosystemseng.2016.04.007 · Cited by 58
Hand segmentation for gesture recognition in EGO-vision
Serra, G., Camurri, M., Baraldi, L., Benedetti, M., & Cucchiara, R. (2013) · doi:10.1145/2505483.2505490 · Cited by 64
“[11].”
Hashmod: A Hashing Method for Scalable 3D Object Detection
Kehl, W., Tombari, F., Navab, N., Ilić, S., & Lepetit, V. (2015) · British Machine Vision Conference (BMVC) · doi:10.5244/c.29.36 · Cited by 64
“While [23, 32] do not consider the pose estimation problem, [14] focuses on this problem only with a discriminatively trained mixture of HOG templates.”
BackgroundRen, H., Wong, A. B., Lian, W., Cheng, W., Zhang, Y., He, J., Liu, Q., Yang, J., Zhang, C. J., Wu, K., et al. (2021) · IEEE Access · doi:10.1109/access.2021.3090215 · Cited by 47
“The Figure 7 shows the calibration curve of MulNet and the estimated probabilities obtained with MulNet by both Isotonic calibration [44] and Sigmoid calibration [45].”
MethodCross-Domain Image Matching with Deep Feature Maps
Kong, B., Supančič, J. S., Ramanan, D., & Fowlkes, C. C. (2019) · International Journal of Computer Vision · doi:10.1007/s11263-018-01143-3 · Cited by 57
“Prior work has explored one-shot object recognition with only a single training example, or “exemplar” (Malisiewicz et al. 2011).”
BackgroundLandmark Classification With Hierarchical Multi-Modal Exemplar Feature
Zhu, L., Shen, J., Jin, H., Xie, L., & Zheng, R. (2015) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2015.2431496 · Cited by 55
“• Exemplar-based approach [21], [22]: This approach has been widely applied in visual object detection, where exemplars are representative regions, and similarities between image instance and exemplars are calculated to generate the similarity feature.”
MethodMining Mid-level Visual Patterns with Deep CNN Activations
Li, Y., Liu, L., Shen, C., & van den Hengel, A. (2016) · International Journal of Computer Vision · doi:10.1007/s11263-016-0945-y · Cited by 64
“this iterative procedure until no more elements can be added into the positive training set. The idea behind this process is using the detection score as a similarity metric, inspired by Exemplar SVM [61,77]. The output of the ensemble merging step is a merged set of mid-level elements and their corresponding detector…”
MethodNYC3DCars: A Dataset of 3D Vehicles in Geographic Context
Matzen, K., & Snavely, N. (2013) · 2013 IEEE International Conference on Computer Vision · doi:10.1109/iccv.2013.99 · Cited by 55
“As in [18], we greedily select the top scoring detection not yet selected or removed, then remove all other detections whose overlap with the selected detection is greater than a threshold (we use a threshold of 0.”
MethodComputer vision approaches for detecting missing barricades
Chian, E. Y. T., Fang, W., Goh, Y. M., & Tian, J. (2021) · Automation in Construction · doi:10.1016/j.autcon.2021.103862 · Cited by 55
Metadata
- Authors
- Malisiewicz, T., Gupta, A., & Efros, A. A.
- Affiliations
- Carnegie Mellon University (US)
- Venue
- 2011 International Conference on Computer Vision
- Publisher
- IEEE
- Published
- 2011-11
- Pages
- 89-96
- OpenAlex
- W1989684337
- Cited by
- 865
- References
- 25
- Citations by year
- 2026: 4 · 2025: 18 · 2024: 16 · 2023: 29 · 2022: 54 · 2021: 52 · 2020: 49 · 2019: 45 · 2018: 65 · 2017: 78 · 2016: 105 · 2015: 148 · 2014: 97 · 2013: 65 · 2012: 36
- FWCI
- 48.23
- Type
- conference paper
- Language
- en
- Open access
- green
- figshare.com
- Licence
- cc-by
- Locations
- IEEE/CVF International Conference on Computer Vision (ICCV) (publishedVersion)
- Figshare (submittedVersion, cc-by)
- KiltHub Repository (submittedVersion)
- KiltHub Repository
- Funders
- National Science Foundation (0905402)
- Multidisciplinary University Research Initiative
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Image and Video Retrieval Techniques
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Visual Attention and Saliency Detection
- Keywords
- object detection, linear SVM classifier, Pascal VOC, scene understanding, part-based models, generalization
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.