Filters








16,756 Hits in 4.3 sec

Learning to Learn from Noisy Web Videos

Serena Yeung, Vignesh Ramanathan, Olga Russakovsky, Liyue Shen, Greg Mori, Li Fei-Fei
2017 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
A promising way to address this is to leverage noisy data from web queries to learn new actions, using semi-supervised or "webly-supervised" approaches.  ...  In this work, we instead propose a reinforcement learning-based formulation for selecting the right examples for training a classifier from noisy web search results.  ...  Acknowledgments Our work is supported by an ONR MURI grant and a hardware donation from NVIDIA.  ... 
doi:10.1109/cvpr.2017.788 dblp:conf/cvpr/YeungRRSMF17 fatcat:zlwutlx7bvcjlniwbq4lmiykxm

Learning to Learn from Noisy Web Videos [article]

Serena Yeung, Vignesh Ramanathan, Olga Russakovsky, Liyue Shen, Greg Mori, Li Fei-Fei
2017 arXiv   pre-print
A promising way to address this is to leverage noisy data from web queries to learn new actions, using semi-supervised or "webly-supervised" approaches.  ...  In this work, we instead propose a reinforcement learning-based formulation for selecting the right examples for training a classifier from noisy web search results.  ...  Acknowledgments Our work is supported by an ONR MURI grant and a hardware donation from NVIDIA.  ... 
arXiv:1706.02884v1 fatcat:lfbm64o6kfgktpucbtvgzrtxvm

Exploiting Multi-modal Curriculum in Noisy Web Data for Large-scale Concept Learning [article]

Junwei Liang, Lu Jiang, Deyu Meng, Alexander Hauptmann
2016 arXiv   pre-print
WELL introduces a number of novel multi-modal approaches to incorporate meaningful prior knowledge called curriculum from the noisy web videos.  ...  Learning video concept detectors automatically from the big but noisy web data with no additional manual annotations is a novel but challenging area in the multimedia and the machine learning community  ...  To address the problem of learning detectors from the big web data, in this paper, we utilize multi-modal information to harness prior knowledge from the web video without any manual efforts.  ... 
arXiv:1607.04780v1 fatcat:ux5a6xqimfcuphi5geusnuur2i

Robust Semantic Video Indexing by Harvesting Web Images [chapter]

Yang Yang, Zheng-Jun Zha, Heng Tao Shen, Tat-Seng Chua
2013 Lecture Notes in Computer Science  
We then develop a robust image-to-video indexing approach to learn reliable classifiers from a limited number of training videos together with abundant user-tagged images.  ...  This paper proposes a robust semantic video indexing framework, which exploits user-tagged web images to assist learning robust semantic video indexing classifiers.  ...  The objective is to develop an effective semantic video indexing approach, which can learn a robust classifier from the limited training videos Z together with abundant user-tagged web images X .  ... 
doi:10.1007/978-3-642-35725-1_7 fatcat:xhynhwch7zdprgwxkd3bjdgg5y

NoisyActions2M: A Multimedia Dataset for Video Understanding from Noisy Labels [article]

Mohit Sharma, Raj Patra, Harshal Desai, Shruti Vyas, Yogesh Rawat, Rajiv Ratn Shah
2021 arXiv   pre-print
In this work, we explore the use of user-generated freely available labels from web videos for video understanding.  ...  We present this as a benchmark dataset in noisy learning for video understanding. The dataset, code, and trained models will be publicly available for future research.  ...  Future Directions This dataset intends to set a standard benchmark to learn from noisy web data in various multimedia tasks.  ... 
arXiv:2110.06827v1 fatcat:syq7w426fzfslnj7rtmcuz24ba

You Lead, We Exceed: Labor-Free Video Concept Learning by Jointly Exploiting Web Videos and Images

Chuang Gan, Ting Yao, Kuiyuan Yang, Yi Yang, Tao Mei
2016 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
As a result, a valid question is how Web images and videos interact for video concept learning.  ...  A plausible solution for training data collection is by sampling from the vast quantities of images and videos on the Web.  ...  To the best of our knowledge, there are no previous works exploring how to obtain reasonable results using noisy Web data. Learning from Web Data.  ... 
doi:10.1109/cvpr.2016.106 dblp:conf/cvpr/GanYYYM16 fatcat:mjwiwwzjcrfjpdytojrsn3peay

How Unlabeled Web Videos Help Complex Event Detection?

Huan Liu, Qinghua Zheng, Minnan Luo, Dingwen Zhang, Xiaojun Chang, Cheng Deng
2017 Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence  
In this paper, we propose a new robust dictionary learning framework for complex event detection, which is able to handle both labeled and easy-to-get unlabeled web videos by sharing the same dictionary  ...  By employing the lq-norm based loss jointly with the structured sparsity based regularization, our model shows strong robustness against the substantial noisy and outlier videos from open source.  ...  We summarize our contributions as follows: • Instead of picking and annotating web videos manually, we pioneer to employ the easy-to-get unlabeled web videos from open source to enhance the performance  ... 
doi:10.24963/ijcai.2017/564 dblp:conf/ijcai/LiuZLZCD17 fatcat:y2q6pl7km5blzlyvc7pt374pba

Learning Actions from the Identity in the Web

Khawla Hussein Ali, Tianjiang Wang
2014 Journal of Computer and Communications  
The idea is to use images collected from the web to learn representations of actions related with identity, use this knowledge to automatically annotate identity in videos.  ...  We present the simple experimental evidence that using action images related with identity collected from the web, annotating identity is possible.  ...  Acknowledgements We would like to thank the anonymous reviewers for their constructive comments and suggestions that help to improve the quality of this manuscript.  ... 
doi:10.4236/jcc.2014.29008 fatcat:xoitz4jidje57fajvr2unnbjjy

Learning facial attributes by crowdsourcing in social media

Yan-Ying Chen, Winston H. Hsu, Hong-Yuan Mark Liao
2011 Proceedings of the 20th international conference companion on World wide web - WWW '11  
., carry rich information for locating designated persons and profiling the communities from image/video collections (e.g., surveillance videos or photo albums).  ...  Our work can (1) automatically extract training images from the semantic-consistent user groups and (2) filter out noisy training photos by multiple mid-level features (by voting).  ...  EXPERIMENTS AND DISCUSSIONS Tackling noisy training photos -We compare our approach with (a) learning by manual annotation [2] and (b) learning by noisy images from the web [1] .  ... 
doi:10.1145/1963192.1963206 dblp:conf/www/ChenHL11 fatcat:7hzmm6byqfea5mwfvnrdt3vo6e

Webly-Supervised Video Recognition by Mutually Voting for Relevant Web Images and Web Video Frames [chapter]

Chuang Gan, Chen Sun, Lixin Duan, Boqing Gong
2016 Lecture Notes in Computer Science  
An alternative and cheap solution is to draw from the large-scale images and videos from the Web.  ...  A question thus naturally arises: to what extent can such noisy Web images and videos be utilized for labeling-free video recognition?  ...  It remains unclear whether they can also obtain reasonable video recognition results using noisy Web data. Learning from weakly-labeled Web Data Web data is inherently noisy.  ... 
doi:10.1007/978-3-319-46487-9_52 fatcat:cw7xhwzoafayxehfy3owtdddwm

Localizing web videos using social images

Liujuan Cao, Xian-Ming Liu, Wei Liu, Rongrong Ji, Thomas Huang
2015 Information Sciences  
In doing so, a novel transfer learning algorithm is proposed to align the landmark prototypes across both domains of images and video frames, leading to a reliable prediction of the geo-locations of web  ...  A group of experiments are carried out on two datasets which collect Flickr images and YouTube videos crawled from the Web.  ...  In this paper, we propose to tackle this challenge from a novel transfer learning perspective, i.e., transferring an accurate and easy-to-learn video geo-tagging model from the image domain.  ... 
doi:10.1016/j.ins.2014.08.017 fatcat:znad7of3gjaqpnw26kowtltxni

Multimodal Co-Training for Selecting Good Examples from Webly Labeled Video [article]

Ryota Hinami, Junwei Liang, Shin'ichi Satoh, Alexander Hauptmann
2018 arXiv   pre-print
We tackle the problem of learning concept classifiers from videos on the web without using manually labeled data.  ...  The main challenge is therefore how to select good examples from noisy training data.  ...  The main challenge is therefore how to select good examples from noisy training data. Liang et al. [26] proposed an approach to learning concept classifiers from web videos with noisy labels.  ... 
arXiv:1804.06057v1 fatcat:z6ir5vtylfeupiili6yeotf6ym

Attention Transfer from Web Images for Video Recognition

Junnan Li, Yongkang Wong, Qi Zhao, Mohan S. Kankanhalli
2017 Proceedings of the 2017 ACM on Multimedia Conference - MM '17  
In this work, we propose a novel approach to transfer knowledge from image domain to video domain.  ...  To harness the rich and highly diverse set of Web images, a scalable approach is to crawl these images to train deep learning based classifier, such as Convolutional Neural Networks (CNN).  ...  Learning from Web Data To harness the information from large-scale Web images, several works use Web images as auxiliary training data for video recognition [10, 11, 23, 37] . Ma et al.  ... 
doi:10.1145/3123266.3123432 dblp:conf/mm/LiWZK17 fatcat:kb7o55xp5fhgpdmzzxvdoryp7u

Attention Transfer from Web Images for Video Recognition [article]

Junnan Li, Yongkang Wong, Qi Zhao, Mohan Kankanhalli
2017 arXiv   pre-print
In this work, we propose a novel approach to transfer knowledge from image domain to video domain.  ...  To harness the rich and highly diverse set of Web images, a scalable approach is to crawl these images to train deep learning based classifier, such as Convolutional Neural Networks (CNN).  ...  Learning from Web Data To harness the information from large-scale Web images, several works use Web images as auxiliary training data for video recognition [10, 11, 23, 37] . Ma et al.  ... 
arXiv:1708.00973v1 fatcat:c3qvuqxzr5a6vcz7wub5yljqii

Learning without prejudice: Avoiding bias in webly-supervised action recognition

Christian Rupprecht, Ansh Kapil, Nan Liu, Lamberto Ballan, Federico Tombari
2017 Computer Vision and Image Understanding  
The key idea is that models such as CNNs can be learned from the noisy visual data available on the web.  ...  One of the main problems in webly-supervised learning is cleaning the noisy labeled data from the web.  ...  Recently, some works have also proposed to learn CNNs [11, 12] and visual concepts [10, 40, 41] from noisy web data.  ... 
doi:10.1016/j.cviu.2017.08.006 fatcat:n4c65lh67jemtdxmt2nf5ivz2y
« Previous Showing results 1 — 15 out of 16,756 results