3,781 Hits in 4.2 sec

Semantic video classification by integrating unlabeled samples for classifier training

Jianping Fan, Hangzai Luo
2004 Proceedings of the 27th annual international conference on Research and development in information retrieval - SIGIR '04  
To address this problem, we have proposed a semi-supervised framework to achieve incremental classifier training by integrating a limited number of labeled samples with a large number of unlabeled samples  ...  However, most existing techniques for classifier training require a large number of hand-labeled samples to learn correctly.  ...  Integrating the unlabeled samples for classifier training not only dramatically reduces the cost for labeling sufficient samples required for accurate classifier training but also increases the classifier  ... 
doi:10.1145/1008992.1009136 dblp:conf/sigir/FanL04 fatcat:w42d72373fc5daqiph6ur4zwia

Semantic video classification and feature subset selection under context and concept uncertainty

Jianping Fan, Hangzai Luo, Jing Xiao, Lide Wu
2004 Proceedings of the 2004 joint ACM/IEEE conference on Digital libraries - JCDL '04  
adaptive EM algorithm to integrate the unlabeled samples to enable incremental classifier training and address the problem of context uncertainty; (d) Proposing a cost-sensitive video classification technique  ...  To address the problems of context and concept uncertainty, we have proposed a novel framework to achieve incremental classifier training by integrating a limited number of labeled samples with a large  ...  by integrating a limited number of labeled samples with a large number of unlabeled samples for classifier training.  ... 
doi:10.1145/996350.996395 dblp:conf/jcdl/FanLXW04 fatcat:44blbmqn5nbhffxol6ih7ex7x4

Incorporating feature hierarchy and boosting to achieve more effective classifier training and concept-oriented video summarization and skimming

Hangzai Luo, Yuli Gao, Xiangyang Xue, Jinye Peng, Jianping Fan
2008 ACM Transactions on Multimedia Computing, Communications, and Applications (TOMCCAP)  
In addition, the unlabeled samples are integrated to reduce the human efforts on labeling large amount of training samples.  ...  For online medical education purposes, we have developed a novel scheme to incorporate the results of semantic video classification to select the most representative video shots for generating concept-oriented  ...  Klara Nahrstedt for her comments and handling the review process of our paper.  ... 
doi:10.1145/1324287.1324288 fatcat:gu7ti4jopzd5dagwjzk3flcow4

Semantic video classification by integrating flexible mixture model with adaptive EM algorithm

Jianping Fan, Hangzai Luo, Xiaodong Lin
2003 Proceedings of the 5th ACM SIGMM international workshop on Multimedia information retrieval - MIR '03  
Since more effective video content characterization framework has been integrated with an adaptive EM algorithm for video classification, our semantic video classifier has improved the classification accuracy  ...  For skin classification, its accuracy is close to 95.5%. For semantic surgical video classification, it achieves overall ≈ 84.6% accuracy.  ...  Our adaptive EM algorithm is very attractive for integrating large-scale unlabeled training samples with the limited number of labeled training samples to obtain a good classifier because the optimal number  ... 
doi:10.1145/973264.973267 dblp:conf/mir/FanLL03 fatcat:7lnkxza4uraq5b7gyrp2bedw7y

A novel approach for privacy-preserving video sharing

Jianping Fan, Hangzai Luo, Mohand-Said Hacid, Elisa Bertino
2005 Proceedings of the 14th ACM international conference on Information and knowledge management - CIKM '05  
transmission and reliably limit the privacy breaches by determining the optimal size of blurred test samples for classifier validation.  ...  To prevent the statistical inferences from video collections, we have developed a distributed framework for privacy-preserving classifier training, which is able to significantly reduce the costs of data  ...  After the weak classifier for the given semantic video concept Cj is learned from a limited number of available labeled samples, the Bayesian framework is used to achieve "soft" classification of unlabeled  ... 
doi:10.1145/1099554.1099711 dblp:conf/cikm/FanLHB05 fatcat:bhxvwhvhfvc4nil3ztouc5dxw4

ClassView : Hierarchical Video Shot Classification, Indexing, and Accessing

J. Fan, A.K. Elmagarmid, X. Zhu, W.G. Aref, L. Wu
2004 IEEE transactions on multimedia  
The hierarchical tree structure of our video database indexing scheme is determined by the domain-dependent concept hierarchy which is also used for video classification.  ...  The Expectation-Maximization (EM) algorithm is also used to determine the classification rule for each visual concept node in the classifier. 2) A hierarchical video database indexing and summary presentation  ...  ACKNOWLEDGMENT The authors thank the reviewers for their useful comments and suggestions to make this paper more readable. They also  ... 
doi:10.1109/tmm.2003.819583 fatcat:4m3bqkdn7rfpnfifit7ieoihvi

TNT: Text-Conditioned Network with Transductive Inference for Few-Shot Video Classification [article]

Andrés Villa, Juan-Manuel Perez-Rua, Victor Escorcia, Vladimir Araujo, Juan Carlos Niebles, Alvaro Soto
2021 arXiv   pre-print
In this paper, we propose to leverage these human-provided textual descriptions as privileged information when training a few-shot video classification model.  ...  Recently, few-shot video classification has received an increasing interest.  ...  Acknowledgements This work was supported in part by the Millennium Institute for Foundational Research on Data (IMFD).  ... 
arXiv:2106.11173v2 fatcat:dgrm6xxutrharom74tcsylfmia

Concept-Oriented Indexing of Video Databases: Toward Semantic Sensitive Retrieval and Browsing

J. Fan, H. Luo, A.K. Elmagarmid
2004 IEEE Transactions on Image Processing  
by using flexible mixture model to bridge the semantic gap; (3) A novel semantic video classifier training framework by integrating feature selection, parameter estimation, and model selection seamlessly  ...  ; (c) Semantic video classification; (d) Concept-oriented video database indexing and access.  ...  Our adaptive EM algorithm is very attractive for integrating largescale unlabeled training samples with the limited number of labeled training samples to obtain a good classifier because the optimal number  ... 
doi:10.1109/tip.2004.827232 pmid:15648863 fatcat:pevnxl3cfzb63fxyqbz3nssgti

A pseudo relevance feedback based cross domain video concept detection

Xu Shaoxi, Yang Jing, Tang Sheng, Zhang Yong-Dong
2011 Proceedings of the Third International Conference on Internet Multimedia Computing and Service - ICIMCS '11  
Then, these pseudo samples are integrated into the process of Tradboost based cross domain transfer learning to make the best use of semantic information generalized by existing source models.  ...  Due to the mismatch of data distribution between training and testing data set, the issue of semantic gap in the field of video concept detection becomes more and more serious.  ...  Then, these pseudo samples are integrated into the process of Tradboost based cross domain transfer learning to make the best use of semantic information generalized by existing source models.  ... 
doi:10.1145/2043674.2043681 dblp:conf/icimcs/XuYTZ11 fatcat:aqk657mghfc4jddvq32f7gm7ra

Semi-supervised learning for semantic video retrieval

Ralph Ewerth, Bernd Freisleben
2007 Proceedings of the 6th ACM international conference on Image and video retrieval - CIVR '07  
Adaboost and Support Vector Machines (SVM) are incorporated for feature selection and ensemble classification. Finally, the newly trained classifiers and the initial model form an ensemble.  ...  Then, two additional classifiers are trained on the automatically labeled data of this video.  ...  Training samples which are misclassified by the current classification model are reweighted such that they have more impact in the next training round for the next "weak classifier".  ... 
doi:10.1145/1282280.1282308 dblp:conf/civr/EwerthF07 fatcat:t25rzf63ovf2bmlon6odiundd4

Cross-domain structural model for video event annotation via web images

Han Wang, Xiabi Liu, Xinxiao Wu, Yunde Jia
2014 Multimedia tools and applications  
In this paper, we try to learn models for video event annotation by leveraging abundant Web images which contains a rich source of information with many events taken under various conditions and roughly  ...  But it is extremely time consuming and labor expensive to collect a large amount of required labeled videos for modeling events under various circumstances.  ...  Acknowledgments This work was partially supported by National Natural Science Foundation of China (Grant no. 60973059, 81171407) and Program for New Century Excellent Talents in University of China (Grant  ... 
doi:10.1007/s11042-014-2175-z fatcat:qn6zq5zc3jgovnfzt2yw4c45bu

Mining Semantic Context Information for Intelligent Video Surveillance of Traffic Scenes

Tianzhu Zhang, Si Liu, Changsheng Xu, Hanqing Lu
2013 IEEE Transactions on Industrial Informatics  
By means of object-specific context information, a cotrained classifier, which takes advantage of the multiview information of objects and reduces the number of labeling training samples, is learned to  ...  classify objects into pedestrians or vehicles with high object classification performance.  ...  OUR CLASSIFIER IS BEST BECAUSE IT FUSES MULTIPLE FEATURES AND ENLARGES TRAINING SET FROM UNLABELED SAMPLES Fig. 9.  ... 
doi:10.1109/tii.2012.2218251 fatcat:viszu3ig2ncp7ntwpqoufe3lcm

Online multi-label active annotation

Xian-Sheng Hua, Guo-Jun Qi
2008 Proceeding of the 16th ACM international conference on Multimedia - MM '08  
Large-scale unlabeled video samples are assumed to arrive consecutively in batches with an initial pre-labeled training set, based on which a preliminary multi-label classifier is built.  ...  And then an online learner updates the original classifier by taking the newly labeled sample-label pairs into consideration. This process repeats until all data are arrived.  ...  Active Learning for Video Annotation Active learning is one of the widely-used approaches in image and video classification, as it can significantly reduce human cost in labeling training samples [5]  ... 
doi:10.1145/1459359.1459379 dblp:conf/mm/HuaQ08 fatcat:va4bdm5vvjehtnp7bfxd2bqhhe

Adaptive Learning for Celebrity Identification With Video Context

Chao Xiong, Guangyu Gao, Zhengjun Zha, Shuicheng Yan, Huadong Ma, Tae-Kyun Kim
2014 IEEE transactions on multimedia  
More specifically, given a few static images and vast face videos, an initial weak classifier is trained and gradually evolves by iteratively promoting the confident tracks into the "labeled" set.  ...  This learning theme may suffer from semantic drifting caused by errors in selecting the confident tracks.  ...  Suppose we are given in total training samples of individuals, which include labeled samples and unlabeled samples (video tracks), i.e., .  ... 
doi:10.1109/tmm.2014.2316475 fatcat:jazbdudshnbsvkpo65hbdcpbnu

Learning semantic scene models by object classification and trajectory clustering

Tianzhu Zhang, Hanqing Lu, S.Z. Li
2009 2009 IEEE Conference on Computer Vision and Pattern Recognition  
In this framework, the detected moving objects are first classified as pedestrians or vehicles via a co-trained classifier which takes advantage of the multiview information of objects.  ...  However, due to the diversity of moving objects' category and their motion patterns, developing robust semantic scene models for activity analysis remains a challenging problem in traffic scenarios.  ...  Acknowledgements This work was supported by the following funding resources: National Natural Science Foundation Project #60833006, #60835002, #60518002.  ... 
doi:10.1109/cvprw.2009.5206809 fatcat:lmgoy3632jdkpnicayzx2beura
« Previous Showing results 1 — 15 out of 3,781 results