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 301–400 of 873
On Tilted Losses in Machine Learning: Theory and ApplicationsPreprint
Tian, L., Beirami, A., Sanjabi, M., & Smith, V. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2109.06141 · Cited by 29
“, 2018), performing classification with imbalanced data (Lin et al., 2017; Malisiewicz et al., 2011), ensuring that subgroups within a population are treated fairly (Hashimoto et al.”
BackgroundMethodInfluentialRobot In a Room: Toward Perfect Object Recognition in Closed EnvironmentsPreprint
Song, S., Zhang, L., & Xiao, J. (2015) · arXiv (Cornell University) · doi:10.48550/arxiv.1507.02703 · Cited by 32
“For category-level recognition, the state-of-the-art object detectors are [12, 8, 24], and [36, 17] for RGB-D images.”
BackgroundSense discovery via co-clustering on images and text
Chen, X., Ritter, A., Gupta, A., & Mitchell, T. M. (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.2015.7299167 · Cited by 25
“For modeling the smoothness term, we compute the similarity (∆(x2i , x 2 j )) between two feature vectors x 2 i and x 2 j using the dot product on whitened HOG feature [21, 29].”
MethodSelf-citationLeveraging local neighborhood topology for large scale person re-identification
Karaman, S., Lisanti, G., Bagdanov, A. D., & Bimbo, A. D. (2014) · Pattern Recognition · doi:10.1016/j.patcog.2014.06.003 · Cited by 25
“Intuitively, the unary cost defines a distribution over labels (identities) at each node, while the smoothness penalty should encourage that nodes corresponding to similar observations take similar labels.”
BackgroundEnsemble Partitioning for Unsupervised Image Categorization
Dai, D., Prasad, M., Leistner, C., & Gool, L. V. (2012) · Lecture notes in computer science · doi:10.1007/978-3-642-33712-3_35 · Cited by 25
“Additionally, they have also been shown to generalize well even when trained only on a few training samples [28], which is an important characteristic as we keep the size of the individual WT sets relatively small.”
A Gaussian Approximation of Feature Space for Fast Image Similarity
Gharbi, M., Malisiewicz, T., Paris, S., & Durand, F. (2012) · DSpace@MIT (Massachusetts Institute of Technology) · Cited by 28
Self-citationMobile Devices as an Infrastructure: A Survey of Opportunistic Sensing Technology
Higuchi, T., Yamaguchi, H., & Higashino, T. (2015) · Journal of Information Processing · doi:10.2197/ipsjjip.23.94 · Cited by 26
“As well as human bodies, other objects or scenes in general can be recognized once their visual model is constructed [63], [64], [65].”
BackgroundTraining Deformable Object Models for Human Detection Based on Alignment and Clustering
Drayer, B., & Brox, T. (2014) · Lecture notes in computer science · doi:10.1007/978-3-319-10602-1_27 · Cited by 25
“An extreme variant, where each training sample defines its own mixture component, was proposed in [18].”
BackgroundUnsupervised underwater fish detection fusing flow and objectiveness
Zhang, D., Kopanas, G., Desai, C., Chai, S. M., & Piacentino, M. (2016) · IEEE/CVF Winter Conference on Applications of Computer Vision Workshops (WACVW) · doi:10.1109/wacvw.2016.7470121 · Cited by 25
Part Discovery from Partial Correspondence
Maji, S., & Shakhnarovich, G. (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.2013.125 · Cited by 23
“Given a sampled “seed”, we initialize the model by training the HOG filter w(0) to separate the seed patch from a set of background patches; this step resembles the exemplar-SVM of [16].”
MethodExemplar-AMMs: Recognizing Crowd Movements From Pedestrian Trajectories
Liu, W., Lau, R. W. H., Wang, X., & Manocha, D. (2016) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2016.2598091 · Cited by 24
Digital Assyriology—Advances in Visual Cuneiform Analysis
Bogacz, B., & Mara, H. (2022) · Journal on Computing and Cultural Heritage · doi:10.1145/3491239 · Cited by 24
“An approach similar to exemplar SVMs [63].”
Underwater Object Detection Based on Enhanced YOLO
Wang, X., Jiang, Z., Xia, Z., & Feng, X. (2022) · International Conference on Image Processing and Media Computing (ICIPMC) · doi:10.1109/icipmc55686.2022.00012 · Cited by 26
Mixture Component Identification and Learning for Visual Recognition
Aghazadeh, O., Azizpour, H., Sullivan, J., & Carlsson, S. (2012) · Lecture notes in computer science · doi:10.1007/978-3-642-33783-3_9 · Cited by 29
“Current approaches in the second direction include the clustering according to the Aspect Ratio of the bounding boxes [1], a combination of HOG and AR similarity [12] and the recent ensemble of exemplar SVM approach [7] which essentially treats each positive sample as a mixture component.”
BackgroundMethodResultInfluentialOf Gods and Goats: Weakly Supervised Learning of Figurative Art
Crowley, E. J., & Zisserman, A. (2013) · British Machine Vision Conference (BMVC) · doi:10.5244/c.27.39 · Cited by 24
“In more detail, the candidate region is used to train a HOG [9] sliding window classifier (in the manner of an exemplar SVM [24] using one positive sample) but implemented here using the more efficient LDA training of [17] which does not require mining for hard negatives.”
MethodScene Parsing with Object Instance Inference Using Regions and Per-exemplar Detectors
Tighe, J., Niethammer, M., & Lazebnik, S. (2014) · International Journal of Computer Vision · doi:10.1007/s11263-014-0778-5 · Cited by 26
Extreme Value Theory-Based Methods for Visual Recognition
Scheirer, W. J. (2017) · Synthesis lectures on computer vision · doi:10.1007/978-3-031-01817-6 · Cited by 25
Recovering 3D existing-conditions of indoor structures from spherical images
Pintore, G., Pintus, R., Ganovelli, F., Scopigno, R., & Gobbetti, E. (2018) · Computers & Graphics · doi:10.1016/j.cag.2018.09.013 · Cited by 24
“Such detector, compared to a standard nearest-neighbor scheme, can be trained in a high discriminative way [16].”
MethodRectified Meta-learning from Noisy Labels for Robust Image-based Plant Disease Classification
Zhai, D., Shi, R., Jiang, J., & Liu, X. (2022) · ACM Transactions on Multimedia Computing Communications and Applications · doi:10.1145/3472809 · Cited by 25
“Representative methods include focal loss [27] and hard negative mining [29].”
MethodPower SVM: Generalization with exemplar classification uncertainty
Zhang, W., Yu, S. X., & Teng, S. (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.2012.6247921 · Cited by 28
Why resampling outperforms reweighting for correcting sampling bias with stochastic gradientsPreprint
An, J., Ying, L., & Zhu, Y. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2009.13447 · Cited by 31
“…there are many well-developed resampling (Mani & Zhang, 2003; He & Garcia, 2009; Maciejewski & Stefanowski, 2011) and reweighting (Kumar et al., 2010; Malisiewicz et al., 2011; Chang et al., 2017) techniques, we only focus on the reweighting approaches that do not change the optimization problem.”
BackgroundMethodSCALPEL: Segmentation Cascades with Localized Priors and Efficient Learning
Weiss, D. A., & Taskar, B. (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.2013.265 · Cited by 23
“Classspecific matching typically requires detecting objects and then adapting a segmentation mask to the detected box from either a single training exemplar [14] or cluster of exemplars [10]; alternatively, fragments of the test image can be matched with fragments of objects of known shape and class [4, 3].”
BackgroundRobust Visual Tracking Using Exemplar-Based Detectors
Gao, C., Chen, F., Yu, J.-G., Huang, R., & Sang, N. (2015) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2015.2513700 · Cited by 25
“, object detection [12], [35], image retrieval [36], midlevel representation discovery for scene classification [37], and action classification [38], [39].”
BackgroundInfluentialEnhanced Deep Learning Framework for Fine-Grained Segmentation of Fashion and Apparel
Usmani, U. A., Happonen, A., & Watada, J. (2022) · Lecture notes in networks and systems · doi:10.1007/978-3-031-10464-0_3 · Cited by 23
From Subcategories to Visual Composites: A Multi-level Framework for Object Detection
Lan, T., Raptis, M., Sigal, L., & Mori, G. (2013) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2013.53 · Cited by 23
“In this work, we use exemplar SVM to discover an initial pool of subcategories which we then refine, through merging, and train part-based models for each resulting subcategory.”
BackgroundMethodInfluentialData-Fusion Techniques for Open-Set Recognition Problems
Córdova, M., Júnior, P. R. M., Rocha, A. D. R., & Torres, R. D. S. (2018) · IEEE Access · doi:10.1109/access.2018.2824240 · Cited by 26
“In closed-set setups, information fusion approaches have often led to good classification results [18], [25], [26], [52], [53].”
BackgroundWriter identification for historical handwritten documents using a single feature extraction method
Chammas, M., Makhoul, A., & Demerjian, J. (2020) · International Conference on Machine Learning and Applications · doi:10.1109/icmla51294.2020.00010 · Cited by 23
“Than we implemented Exemplar-SVM (ESVM) using a linear kernel [18] [12], as a result we reached an improved precision and accuracy.”
MethodUnsupervised behaviour analysis and magnification (uBAM) using deep learning
Brattoli, B., Büchler, U., Dorkenwald, M., Reiser, P., Filli, L., Helmchen, F., Wahl, A.-S., & Ommer, B. (2021) · Nature Machine Intelligence · doi:10.1038/s42256-021-00326-x · Cited by 25
Zhang, C., Liu, J., Tian, Q., Liang, C., & Huang, Q. (2013) · Neurocomputing · doi:10.1016/j.neucom.2012.07.056 · Cited by 22
“To alleviate this problem, the use of exemplar training data is also used by researchers [24] [25].”
MethodScalable multi-class geospatial object detection in high-spatial-resolution remote sensing images
Cheng, G., Han, J., Zhou, P., & Guo, L. (2014) · IEEE International Geoscience and Remote Sensing Symposium (IGARSS) · doi:10.1109/igarss.2014.6946975 · Cited by 22
“To evaluate the presented method, we compared it with some state-of-the-art object detection methods [12, 17, 18].”
BackgroundMethodInfluentialEfficient object detection using convolutional neural network-based hierarchical feature modeling
Lee, B., Erdenee, E., Jin, S., & Rhee, P. K. (2016) · Signal Image and Video Processing · doi:10.1007/s11760-016-0962-x · Cited by 23
“However, most of them only consider intra-class variations [10,11] and do not investigate inter-class ambiguity, even though performance can be improved further by considering ambiguity between inter-classes.”
BackgroundObject Detection Based on Convolutional Neural Network
钟, 文. (2020) · Software Engineering and Applications · doi:10.12677/sea.2020.91005 · Cited by 22
SMSPL: Robust Multimodal Approach to Integrative Analysis of Multiomics Data
Yang, Z., Wu, N., Liang, Y. Q., Zhang, H., & Ren, Y. (2020) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2020.3006240 · Cited by 22
“This type of method includes boosting algorithms (e.g., AdaBoost [18]), hard negative mining [27], and focal loss [19].”
MethodReal‐time transmission tower detection from video based on a feature descriptor
Cerón, A., Mondragon, I. F., & Prieto, F. (2016) · IET Computer Vision · doi:10.1049/iet-cvi.2015.0477 · Cited by 22
“Another interesting study [4], which used a rigid template of HOG descriptors and exemplar SVM had a good performance, as compared with the previous approach.”
ResultWe Don’t Need No Bounding-Boxes: Training Object Class Detectors Using Only Human Verification
Papadopoulos, D. P., Uijlings, J. R. R., Keller, F., & Ferrari, V. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.99 · Cited by 21
“Training a detector typically requires a large set of images in which objects have been manually annotated with bounding-boxes [10, 16, 18, 19, 20, 34, 53, 57, 62].”
BackgroundRandomized Max-Margin Compositions for Visual Recognition
Eigenstetter, A., Takami, M., & Ommer, B. (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.2014.459 · Cited by 21
“Moreover, the individual contributions of compositions and their parts are evaluated in separate experiments that demonstrate their potential.”
BackgroundAssociative Embeddings for Large-Scale Knowledge Transfer with Self-Assessment
Vezhnevets, A. S., & Ferrari, V. (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.2014.255 · Cited by 23
“Moreover, E-SVMs that are learnt from similar exemplars coming from the same aspect of object’s appearance (e.g., gorilla’s face, close up) will produce correlated outputs uniformly for all windows.”
BackgroundMethodInfluentialZhang, C., Zhu, C., Xu, X., Liu, Y., Xiao, J., & Tillo, T. (2018) · Signal Processing Image Communication · doi:10.1016/j.image.2018.05.006 · Cited by 21
Exemplar-Based Colour Constancy
Joze, H. R. V., & Drew, M. S. (2012) · British Machine Vision Conference (BMVC) · doi:10.5244/c.26.26 · Cited by 23
“Such use of these techniques occurs in a variety of computer vision problems such as shape matching [3], character recognition [2], human pose estimation [23], image parsing [20], scene recognition [31] and object detection [22].”
BackgroundFamily Member Identification from Photo Collections
Dai, Q., Carr, P., Sigal, L., & Hoiem, D. (2015) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv.2015.136 · Cited by 21
“Instead, we employ an iterative approach borrowing ideas from exemplar SVMs [14].”
MethodConvolutional Neural Network-Based Human Identification Using Outer Ear Images
Sinha, H., Manekar, R., Sinha, Y., & Ajmera, P. K. (2018) · Advances in intelligent systems and computing · doi:10.1007/978-981-13-1595-4_56 · Cited by 21
Cross-Domain Forensic Shoeprint Matching
Kong, B., Supančič, J. S., Ramanan, D., & Fowlkes, C. (2017) · Iowa State University Digital Repository (Iowa State University) · Cited by 20
A Probabilistic Formulation for Meta-Weight-Net
Zhao, Q., Shu, J., Yuan, X., Liu, Z., & Meng, D. (2021) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2021.3105104 · Cited by 23
“The other class consists of weighting functions that are monotonically increasing with loss values, including AdaBoost [24], [25] and its variants [38], hard negative mining [10], and focal loss [11].”
BackgroundMethodDeep object detection for waterbird monitoring using aerial imagery
Kabra, K., Xiong, A., Li, W., Luo, M., Lu, W., Yu, T., Yu, J., Singh, D., Garcia, R. V., Tang, M., et al. (2022) · International Conference on Machine Learning and Applications · doi:10.1109/icmla55696.2022.00073 · Cited by 20
Ahmed, E., Cohen, S. D., & Price, B. L. (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.2014.403 · Cited by 20
“Since object detection localizes a object with a bounding box, it could easily be used to provide a bounding box around a desired object to initialize a segmentation process using a method such as Grabcut [20].”
BackgroundMethodSDM-BSM: A fusing depth scheme for human action recognition
Liu, H., Tian, L., Liu, M., & Tang, H. (2015) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2015.7351693 · Cited by 20
“Then a non-linear SVM classifi-er with a homogenous kernel [12] is trained for the obtained feature vectors.”
MethodVisual Object Detection Using Cascades of Binary and One-Class Classifiers
Çevıkalp, H., & Triggs, B. (2017) · International Journal of Computer Vision · doi:10.1007/s11263-016-0986-2 · Cited by 23
“The SVM classifies an example as a positive if and only if w x + b > 0, where w is the SVM’s weight vector and b is its offset.”
MethodInfluentialSun, H., Meng, Z., Tao, P. Y., & Ang, M. H. (2018) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2018.8460535 · Cited by 22
“Row 2 shows the performance of PlacesCNN scene feature [18] with a Linear SVM.”
BackgroundMethodInfluentialMultiresolution approach for multiple human detection using moments and local binary patterns
Nigam, S., & Khare, A. (2014) · Multimedia Tools and Applications · doi:10.1007/s11042-014-1951-0 · Cited by 20
“The major advantage of using RBF kernel is that it can be applied on feature data of almost any dimension [19].”
Modeling spatial layout for scene image understanding via a novel multiscale sum-product network
Yuan, Z., Wang, H., Wang, L., Lü, T., Shivakumara, P., & Tan, C. L. (2016) · Expert Systems with Applications · doi:10.1016/j.eswa.2016.07.015 · Cited by 20
Malagoli, E., & Persio, L. D. (2025) · Mathematics · doi:10.3390/math13060893 · Cited by 20
Moving Object Detection from Moving Camera Image Sequences Using an Inertial Measurement Unit Sensor
Jung, S., Cho, Y., Kim, D., & Chang, M. (2019) · Applied Sciences · doi:10.3390/app10010268 · Cited by 21
Human-autonomy sensor fusion for rapid object detection
Robinson, R. M., Lee, H., McCourt, M. J., Marathe, A. R., Kwon, H., Ton, C. T., & Nothwang, W. D. (2015) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2015.7353390 · Cited by 19
“• Exemplar Support Vector Machine (ESVM): ESVM learns a separate classifier for each positive training image using a rigid HOG template, and scores candidate detections based on “distance” to exemplars [24].”
BackgroundMethodReal-Time Hand Posture Recognition for Human-Robot Interaction Tasks
Hernandez-Belmonte, U. H., & Ayala-Ramírez, V. (2016) · Sensors · doi:10.3390/s16010036 · Cited by 20
“[45].”
Behavior Discovery and Alignment of Articulated Object Classes from Unstructured Video
Pero, L. D., Ricco, S., Sukthankar, R., & Ferrari, V. (2016) · International Journal of Computer Vision · doi:10.1007/s11263-016-0939-9 · Cited by 21
“They can also enable annotating large collections with little manual effort via knowledge transfer [20, 43, 44, 51, 76].”
BackgroundCross-scenario clothing retrieval and fine-grained style recognition
Li, Z., Li, Y., Tian, W., Pang, Y., & Liu, Y. (2016) · Proceedings - International Conference on Pattern Recognition/Proceedings/International Conference on Pattern Recognition · doi:10.1109/icpr.2016.7900079 · Cited by 19
“We trained a number of E-SVM classifiers on Fashionista database [20] by some selected clothing examples and one classifier was for one clothing example.”
MethodDataset-Driven Unsupervised Object Discovery for Region-Based Instance Image Retrieval
Zhang, Z., Wang, L., Wang, Y., Zhou, L., Zhang, J., & Chen, F. (2022) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2022.3141433 · Cited by 19
“So we put the idea in [40] to the extreme to construct the “base-detector repository”.”
BackgroundMethodInfluentialEnsemble Deep Learning for Human-Object Interaction Detection
Mansour, A. E., Mohammed, A., Elsayed, H. A. E. A., & Ramly, S. H. E. (2022) · International Mobile, Intelligent, and Ubiquitous Computing Conference (MIUCC) · doi:10.1109/miucc55081.2022.9781774 · Cited by 18
Segmentation over Detection by Coupled Global and Local Sparse Representations
Xia, W., Zheng, S., Feng, J., Cheong, L., & Yan, S. (2012) · Lecture notes in computer science · doi:10.1007/978-3-642-33715-4_48 · Cited by 18
Scene parsing by nonparametric label transfer of content-adaptive windows
Tung, F., & Little, J. J. (2015) · Computer Vision and Image Understanding · doi:10.1016/j.cviu.2015.08.009 · Cited by 19
“Tighe and Lazebnik 21] extended the SuperParsing algorithm with per-exemplar detecors (Exemplar-SVMs [22]).”
BackgroundA scalable architecture for multi-class visual object detection
Advani, S., Tanabe, Y., Irick, K. M., Sampson, J., & Narayanan, V. (2015) · International Conference on Field Programmable Logic and Applications (FPL) · doi:10.1109/fpl.2015.7293961 · Cited by 19
“…do Bin gradient magnitudes based on orientation end for Group cells into overlapping blocks for all blocks do Compute normalization factor end for Apply normalization to each overlapping cell Accumulate values as a feature vector Apply linear SVM at all scales and locations Fuse overlapping…”
BackgroundWeighted Part Context Learning for Visual Tracking
Zhu, G., Wang, J., Zhao, C., & Lu, H. (2015) · IEEE Transactions on Image Processing · doi:10.1109/tip.2015.2479460 · Cited by 19
3D Pose-by-Detection of Vehicles via Discriminatively Reduced Ensembles of Correlation Filters
Movshovitz-Attias, Y., Sheikh, Y. A., Boddeti, V. N., & Wei, Z. (2014) · British Machine Vision Conference (BMVC) · doi:10.5244/c.28.53 · Cited by 20
“Exemplar classifiers [20] are trained using a single positive instance and a large set of negative instances.”
BackgroundFast Template Evaluation with Vector Quantization
Sadeghi, M. A., & Forsyth, D. A. (2013) · neural information processing systems · Cited by 21
Lean histogram of oriented gradients features for effective eye detection
Sharma, R., & Savakis, A. E. (2015) · Journal of Electronic Imaging · doi:10.1117/1.jei.24.6.063007 · Cited by 18
Visual co-occurrence network: using context for large-scale object recognition in retail
Advani, S., Smith, B., Tanabe, Y., Irick, K. M., Cotter, M. J., Sampson, J., & Narayanan, V. (2015) · IEEE/ACM/IFIP Workshop on Embedded Systems for Real-Time Multimedia (ESTIMedia) · doi:10.1109/estimedia.2015.7351774 · Cited by 18
“The vision pipeline consists of a two-tier classification system consisting of HMAX features [20] trained on a single Regularized Least Squares (RLS) classifier and Histogram of Oriented Gradients (HOG) features trained using Exemplar Support Vector Machines (ESVM) [21].”
MethodRobust watch-list screening using dynamic ensembles of SVMs based on multiple face representations
Bashbaghi, S., Granger, É., Sabourin, R., & Bilodeau, G.-A. (2017) · Machine Vision and Applications · doi:10.1007/s00138-016-0820-4 · Cited by 18
“Given the target and non-target data available for design, one-class Support Vector Machine (SVM) and the exemplar SVM (2-class) [37] are considered for the base classifiers.”
BackgroundMethodInfluential3D Visual passcode: Speech-driven 3D facial dynamics for behaviometrics
Zhang, J., & Fisher, R. B. (2019) · Signal Processing · doi:10.1016/j.sigpro.2019.02.025 · Cited by 24
“It also has a similarity to the exemplar-SVM [47], although the E-SVM focuses on intraclass details via an ensemble classifier.”
Fast Training of Pose Detectors in the Fourier Domain
Henriques, J. F., Martins, P., Batista, J., & Proofs, A. (2016) · Cited by 20
Learning to Match Images in Large-Scale Collections
Cao, S., & Snavely, N. (2012) · Lecture notes in computer science · doi:10.1007/978-3-642-33863-2_26 · Cited by 18
“On the other hand, recent work has proposed learning per-image classifiers or similarity functions [27, 28].”
BackgroundTowards Domain Adaptive Vehicle Detection in Satellite Image by Supervised Super-Resolution Transfer
Cao, L., Ji, R., Wang, C., & Li, J. (2016) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v30i1.10166 · Cited by 18
“Therefore, it is a more practical solution to train instance-wised classifiers with instance-specific metrics to conduct a refined nearest neighbor classification (Malisiewicz, Gupta, and Efros 2011).”
MethodInfluentialDomain-Constraint Transfer Coding for Imbalanced Unsupervised Domain Adaptation
Tsai, Y.-H. H., Hou, C.-A., Chen, W.-Y., Yeh, Y.-R., & Wang, Y.-C. F. (2016) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v30i1.10443 · Cited by 17
“Extended form exemplar-SVMs (Malisiewicz et al. 2011), low-rank exemplar SVM (LRE-SVM) (Xu et al.”
MethodWriter Identification and Writer Retrieval Using Vision Transformer for Forensic Documents
Koepf, M., Kleber, F., & Sablatnig, R. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-06555-2_24 · Cited by 17
Hundreds Guide Millions: Adaptive Offline Reinforcement Learning With Expert Guidance
Yang, Q., Wang, S., Zhang, Q., Huang, G. M., & Song, S. (2023) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2023.3293508 · Cited by 21
“Generally speaking, methods in this category either force sample weights to monotonically increase (such as AdaBoost [30], hard example mining [31], and focal loss [32]) or decrease (such as self-paced learning (SPL) [54], active bias [55], and iterative reweighting [56]).”
BackgroundMethodGenerative regularization with latent topics for discriminative object recognition
Rubio, J. C., Eigenstetter, A., & Ommer, B. (2015) · Pattern Recognition · doi:10.1016/j.patcog.2015.06.013 · Cited by 17
“Extending to hundreds of specific part detectors helps to capture peculiar characteristics but due to their specificity, for each object instance different parts will be helpful and others will yield noisy responses that actually impair classification.”
BackgroundMethodInfluentialPredicting Functional Regions on Objects
Desai, C., & Ramanan, D. (2013) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2013.141 · Cited by 17
“Supervised part models: Our top-down models are based on exemplar-based templates [21] and 2D pictorial structures [11].”
MethodTarget Detection Using Sparse Representation With Element and Construction Combination Feature
Liu, H., & Shutao Li, A. (2014) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2014.2343412 · Cited by 17
“To discriminate the target and background, many classification methods have been used, such as support vector machine (SVM) [3], [4], sparse representation (SR) [5], and neural networks [6].”
MethodExemplar codes for facial attributes and tattoo recognition
Wilber, M. J., Rudd, E. M., Heflin, B. C., Lui, Y. M., & Boult, T. E. (2014) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv.2014.6836099 · Cited by 17
“Platt normalization used by the ExemplarEnsemble approach [16] as well as W-score fitting [21] work differently.”
BackgroundMethodInfluentialRandom Geometric Prior Forest for Multiclass Object Segmentation
Liu, X., Song, M., Tao, D., Bu, J., & Chen, C. (2015) · IEEE Transactions on Image Processing · doi:10.1109/tip.2015.2432711 · Cited by 18
Robust individual and holistic features for crowd scene classification
Liu, W., Lau, R. W. H., & Manocha, D. (2016) · Pattern Recognition · doi:10.1016/j.patcog.2016.03.031 · Cited by 17
A deep learning based system for writer identification in handwritten Arabic historical manuscripts
Chammas, M., Makhoul, A., Demerjian, J., & Dannaoui, E. (2022) · Multimedia Tools and Applications · doi:10.1007/s11042-022-12673-x · Cited by 17
Simeth, A., Kumar, A. A., & Plapper, P. (2024) · Journal of Intelligent Manufacturing · doi:10.1007/s10845-024-02411-5 · Cited by 20
“The HOG detector has been an essential foundation of many object detectors (Felzenszwalb et al., 2008, 2010; Malisiewicz et al., 2011) and a large variety of CV applications for many years.”
Detect2Rank: Combining Object Detectors Using Learning to Rank
Karaoğlu, S., Liu, Y., & Gevers, T. (2015) · IEEE Transactions on Image Processing · doi:10.1109/tip.2015.2499702 · Cited by 18
“Additionally, we use the confidence scores of the maximum overlapping neighbors of detections by EES [12] in eq.”
BackgroundMethodInfluentialData-driven exemplar model selection
Misra, I., Shrivastava, A., & Hebert, M. (2014) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv.2014.6836080 · Cited by 16
“Hence, we use Platt calibration [26, 30] (on validation set) to interpret the SVM scores as probabilities.”
BackgroundMethodInfluentialLearning Shared, Discriminative, and Compact Representations for Visual Recognition
Löbel, H., Vidal, R., & Soto, Á. (2015) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2015.2408349 · Cited by 17
Joint crop and tassel segmentation in the wild
Lü, H., Cao, Z., Xiao, Y., Li, Y., & Zhu, Y. (2015) · Chinese Automation Congress · doi:10.1109/cac.2015.7382547 · Cited by 16
“Therefore, the choice of multiple models can greatly reduce the difficulty of learning process [22].”
BackgroundData-Driven Detection of Prominent Objects
Rodríguez-Serrano, J. A., Larlus, D., & Dai, Z. (2015) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2015.2509988 · Cited by 21
“In the first strategy, approaches still perform sliding window searches and each sub-window is used as a query for which the nearest-neighbor similarity is computed on huge databases of images [66] or sub-images [50].”
MethodRecent Object Detection Techniques: A Survey
Diwakar, & Raj, D. (2022) · International Journal of Image Graphics and Signal Processing · doi:10.5815/ijigsp.2022.02.05 · Cited by 21
“Malisiewicz et al.[22] describe an exemplar-based approach in a discriminative framework for object detection based on training a separate linear SVM classifier for every exemplar in the training set (2011).”
Explainable detection of adverse drug reaction with imbalanced data distribution
Wang, J., Yu, L.-C., & Zhang, X. (2022) · PLoS Computational Biology · doi:10.1371/journal.pcbi.1010144 · Cited by 16
“To address the data imbalance issue, several previous works applied either down-sampling [44] or up-sampling strategies [44, 45] to balance the data distribution and exploit the most difficult examples of the minority class.”
MethodSpiking Neural Network Based on Multi-Scale Saliency Fusion for Breast Cancer Detection
Fu, Q., & Dong, H. (2022) · Entropy · doi:10.3390/e24111543 · Cited by 20
“Traditional object detection algorithms are based on manual feature extraction, which slowly improves the detection precision by building complex models and multi-model integration based on basic feature expression [12,13].”
Fast scale-invariant lateral lumbar vertebrae detection and segmentation in X-ray images
Sa, R., Owens, W., Wiegand, R., & Chaudhary, V. (2016) · Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) · doi:10.1109/embc.2016.7590884 · Cited by 16
“Apart from the fast framework, inspired by exemplarSVMs proposed by Malisiewicz [9], we categorize the vertebra into three classes: right-tilted, flat and left-tilted classes instead of using different labels for each vertebra or using one label for all vertebrae.”
MethodMining human movement evolution for complex action recognition
Yi, Y., Cheng, Y., & Xu, C. (2017) · Expert Systems with Applications · doi:10.1016/j.eswa.2017.02.020 · Cited by 16
Web-Scale Training for Face IdentificationPreprint
Taigman, Y., Yang, M., Ranzato, M., & Wolf, L. (2014) · arXiv (Cornell University) · doi:10.48550/arxiv.1406.5266 · Cited by 18
“In [16] multiple SVM classifiers, each based on a single positive sample, are used to construct a powerful descriptor of the learned class.”
MethodIncorporating query-specific feedback into learning-to-rank models
Can, E. F., Croft, W. B., & Manmatha, R. (2014) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/2600428.2609503 · Cited by 15
“Our approach is related in spirit to Malisiewicz et al. [12].”
MethodCapturing relative motion and finding modes for action recognition in the wild
Oshin, O., Gilbert, A., & Bowden, R. (2014) · Computer Vision and Image Understanding · doi:10.1016/j.cviu.2014.04.005 · Cited by 15
“[6] both perform local classification of object category recognition, Zhang uses a k-Nearest Neighbour method to approximate the SVM, while Malisiewicz combine the effectiveness of a discriminative object detector with the explicit correspondence offered by a nearest-neighbour approach.”
BackgroundMethodMultiple Object Tracking in Urban Traffic Scenes with a Multiclass Object Detector
Ooi, H.-L., Bilodeau, G.-A., Saunier, N., & Beaupré, D.-A. (2018) · Lecture notes in computer science · doi:10.1007/978-3-030-03801-4_63 · Cited by 15
Class-Aware Domain Adaptation for Coastal Land Cover Mapping Using Optical Remote Sensing Imagery
Chen, J., Chen, G., Fang, B., Wang, J., & Wang, L. (2021) · IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · doi:10.1109/jstars.2021.3128527 · Cited by 16
See the Difference: Direct Pre-Image Reconstruction and Pose Estimation by Differentiating HOG
Chiu, W.-C., & Fritz, M. (2015) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.61 · Cited by 15
“Most prominently it is a core building block of the widely used Deformable Part Model (DPM) object class detector [7] and exemplar models [16] which both on their own have seen many follow-up approaches.”
MethodAlmazán, J., Fernandez, D., Fornés, A., Lladós, J., & Valveny, E. (2012) · International Conference on Frontiers in Handwriting Recognition · doi:10.1109/icfhr.2012.151 · Cited by 15
“[8] in the sense that they also learn a filter form HOG descriptor using Support Vector Machine (SVM).”
MethodDynamic Scene Classification Using Redundant Spatial Scenelets
Du, L., & Ling, H. (2015) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2015.2466692 · Cited by 15
“, e-SVM detectors [34]) trained from discriminative patches.”
BackgroundMethod
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.