Filters








71,268 Hits in 7.0 sec

Leveraging loosely-tagged images and inter-object correlations for tag recommendation

Yi Shen, Jianping Fan
2010 Proceedings of the international conference on Multimedia - MM '10  
the inter-object correlations for classifier training.  ...  To enhance the discrimination power of a large number of inter-related object classifiers, a multi-task structured SVM algorithm is developed to model the inter-task relatedness more precisely and leverage  ...  correlations and the inter-task relatedness for inter-related classifier training and enhance their discrimination power significantly; (b) it can save the cost for object detection by exploiting the  ... 
doi:10.1145/1873951.1873956 dblp:conf/mm/ShenF10 fatcat:qhuydkizczhrpjlmbuhmwm26fe

Multi-view face and eye detection using discriminant features

Peng Wang, Qiang Ji
2007 Computer Vision and Image Understanding  
Multi-view face detection plays an important role in many applications. This paper presents a statistical learning method to extract features and construct classifiers for multi-view face detection.  ...  The RNDA relaxes Gaussian assumptions of Fisher discriminant analysis (FDA), and it can handle more general class distributions.  ...  The RNDA is more efficient for a high dimensional space and a large training set, which is the case for face detection.  ... 
doi:10.1016/j.cviu.2006.08.008 fatcat:7r4d47mxerfldlthumfb4eaeeu

CALEB: A Conditional Adversarial Learning Framework to Enhance Bot Detection [article]

George Dialektakis, Ilias Dimitriadis, Athena Vakali
2022 arXiv   pre-print
Finally, the use of the AC-GAN Discriminator as a bot detector, has outperformed former ML approaches, showcasing the efficiency of our end to end framework.  ...  on the Conditional Generative Adversarial Network (CGAN) and its extension, Auxiliary Classifier GAN (AC-GAN), to simulate bot evolution by creating realistic synthetic instances of different bot types  ...  Secondly, we can utilize the discriminator of the model for multi-class bot classification, reducing the training overhead that applies when training additional ML classifiers on the synthetic data for  ... 
arXiv:2205.15707v1 fatcat:fvd26crb6jgstkjyrrk3q22k2a

Multi-class Generative Adversarial Nets for Semi-supervised Image Classification [article]

Saman Motamed, Farzad Khalvati
2021 arXiv   pre-print
We propose a modification to the traditional training of GANs that allows for improved multi-class classification in similar classes of images in a semi-supervised learning framework.  ...  With the remarkable ability of GANs in learning the distribution and generating images of a particular class, they can be used for semi-supervised classification tasks.  ...  Acknowledgements This research was funded by Chair in Medical Imaging and Artificial Intelligence funding, a joint Hospital-University Chair between the University of Toronto, The Hospital for Sick Children  ... 
arXiv:2102.06944v2 fatcat:th4bm57kejhkdepi4kogyve444

Novelty Detection with GAN [article]

Mark Kliger, Shachar Fleishman
2018 arXiv   pre-print
We show that a multi-class discriminator trained with a generator that generates samples from a mixture of nominal and novel data distributions is the optimal novelty detector.  ...  The ability of a classifier to recognize unknown inputs is important for many classification-based systems.  ...  This mixture generator generates samples scattered around and in the low-density areas of the data manifold, and this makes a multi-class discriminator a powerful novelty detector.  ... 
arXiv:1802.10560v1 fatcat:iytzxnvsq5bmffm2upz63d73qa

Implementation of a Brain-Computer Interface Based on Three States of Motor Imagery

Yijun Wang, Bo Hong, Xiaorong Gao, Shangkai Gao
2007 IEEE Engineering in Medicine and Biology Society. Conference Proceedings  
Then, the optimized parameters and classifiers were utilized for online control.  ...  In this paper, we implemented a three-class BCI manipulated through imagination of left hand, right hand and foot movements, inducing different spatial patterns of event-related desynchronization (ERD)  ...  Multi-step training and controlling procedures 1) Online feedback training Linear discriminant analysis (LDA) was used to classify the band-pass power features on C3/C4 electrodes referenced to FCz [9  ... 
doi:10.1109/iembs.2007.4353477 pmid:18003143 fatcat:w7dojs3eqnhz7pncurcv4oknea

Is a detector only good for detection?

Quan Yuan, Stan Sclaroff
2009 2009 IEEE 12th International Conference on Computer Vision  
For example, consider face detection by a boosted cascade of detectors followed by face ID recognition via one-vs-all (OVA) classifiers.  ...  on a multi-view vehicle data set.  ...  of within-class classifiers that are invoked (for multi-class classification approaches), with little or no impact on accuracy.  ... 
doi:10.1109/iccv.2009.5459389 dblp:conf/iccv/YuanS09 fatcat:wjejar5r7fhidlwokswey4nwie

A fine-grained approach to scene text script identification [article]

Lluis Gomez, Dimosthenis Karatzas
2016 arXiv   pre-print
In addition, we propose a new public benchmark dataset for the evaluation of joint text detection and script identification in natural scenes.  ...  We detail a novel method for script identification in natural images that combines convolutional features and the Naive-Bayes Nearest Neighbor classifier.  ...  the train dataset accounting for its discriminative power.  ... 
arXiv:1602.07475v1 fatcat:xesy7sxekvcwvdzsuxfg6ygfcu

Fast Human Pose Detection Using Randomized Hierarchical Cascades of Rejectors

Grégory Rogez, Jonathan Rihan, Carlos Orrite-Uruñuela, Philip H. S. Torr
2012 International Journal of Computer Vision  
For each branch of this decision tree, we take advantage of the alignment of training images to build a list of potentially discriminative HOG (Histograms of Orientated Gradients) features.  ...  The resulting multi-class classifier is then used to scan images in a sliding window scheme.  ...  They are also grateful to Dr Srikumar Ramalingam for his contribution on the initial version of this paper.  ... 
doi:10.1007/s11263-012-0516-9 fatcat:dkxykefrnjf65eaecq7vh5bvfq

Scalable multi-class object detection

Nima Razavi, Juergen Gall, Luc Van Gool
2011 CVPR 2011  
To this end, a shared discriminative codebook of feature appearances is jointly trained for all classes and detection is also performed for all classes jointly.  ...  Our method has linear training and sublinear detection complexity in the number of classes.  ...  At each node of the hierarchy, a classifier is trained and combined to a multi-class classifier.  ... 
doi:10.1109/cvpr.2011.5995441 dblp:conf/cvpr/RazaviGG11 fatcat:rhtpq7uexjbk5hgajrrv5pdaha

Fast Multi-View Face Tracking With Pose Estimation

Taner Arsan, Javier Cruz Mota, Julien Meynet, Jean-Philippe Thiran
2008 Zenodo  
However, the GF model keeps improving, as we add more and more features. This shows that the HFs are not discriminant enough for modeling the finer differences between the two classes.  ...  The aim of these classifiers is to reject false positive windows but also to discriminate between different poses. That is why they are trained using the more discriminant Gaussian filters.  ... 
doi:10.5281/zenodo.41075 fatcat:35mww4ev4rdktktrsre65tn3pu

Sequential Max-Margin Event Detectors [chapter]

Dong Huang, Shitong Yao, Yi Wang, Fernando De La Torre
2014 Lecture Notes in Computer Science  
Existing event detection methods rely on one-versus-all or multi-class classifiers that do not scale well to online detection of large number of events.  ...  This approach has two main benefits w.r.t. standard approaches: (1) It provides an efficient solution for early detection of events in the presence of large number of classes, and (2) it is computationally  ...  Unlike standard multi-class approaches, SMMED sequentially discards classes that can be early discriminated from the true class, being more efficient when detecting large number of classes.  ... 
doi:10.1007/978-3-319-10578-9_27 fatcat:g5zjthhzvfeblohke2y2gm6smu

Real-time vision-based infotainment user determination for driver assistance

Shinko Y. Cheng, Mohan M. Trivedi
2008 2008 IEEE Intelligent Vehicles Symposium  
This approach represents an alternative of detecting and tracking the hand movements and then classifying the hands into the respective classes. IEEE Intelligent Vehicles Symposium  ...  In this paper, we develop and evaluate a novel real-time computer vision algorithm to robustly discriminate which of the front-row seat occupants is accessing the infotainment controls.  ...  We would also like to thank the kind assistance from members of the Computer Vision and Robotics Research Laboratory for their hand in data collection.  ... 
doi:10.1109/ivs.2008.4621159 fatcat:yuejbrex7rfu7m52p74meimk3e

A multi-class classification strategy for Fisher scores: Application to signer independent sign language recognition

Oya Aran, Lale Akarun
2010 Pattern Recognition  
We experimentally show that the Fisher scores of one class provide discriminative information for the other classes as well.  ...  The proposed multi-class classification strategy increases the classification accuracy in comparison with the state of the art strategies based on combining binary classifiers.  ...  For a problem of K classes, a single multi-class classifier is trained, using the original class labels.  ... 
doi:10.1016/j.patcog.2009.12.002 fatcat:f4mpgsyadbbozi62czyuvwmxny

Autonomous Audio-Supported Learning of Visual Classifiers for Traffic Monitoring

Horst Bischof, Martin Godec, Christian Leistner, Bernhard Rinner, Andreas Starzacher
2010 IEEE Intelligent Systems  
Our system consists of a robust on-line boosting classifier that allows for continuous learning and concept drift.  ...  In this paper we focus on autonomous visual detection and classification of vehicles.  ...  Since most traffic applications are not only concerned in detecting vehicles but also in discriminating different vehicle classes, we train two different detectors-one for trucks and one for cars.  ... 
doi:10.1109/mis.2010.28 fatcat:usiyq77x3bfpllmrfdy4hqa2eq
« Previous Showing results 1 — 15 out of 71,268 results