Cross-view action recognition via view knowledge transfer

Liu, Jingen; Shah, Mubarak; Kuipers, Benjamin; Savarese, Silvio

doi:10.1109/cvpr.2011.5995729

Cited by 248 publications

(228 citation statements)

References 18 publications

Supporting

Mentioning

226

Contrasting

Order By: Relevance

“…Step 1: Given Ω and {Q g }, find Θ from (12). The key ideas is that for given {Q g }, the optimization problem of (12) is separable, i.e., the columns of Θ, {θ v : v = 1, ..., V }, can be independently estimated.…”

Section: Latent Multitask Learningmentioning

confidence: 99%

“…In this step, we use the gradient descent, following the derivation presented in [10]. The gradient decent of the Lagrangian function of (12) is made possible in [10] by relaxing the integer regularization in (12) to g Θ Q g 2 * . For binary solutions of Q g the relaxed regularization is equivalent to the original one.…”

Section: Latent Multitask Learningmentioning

confidence: 99%

“…They seek to extend knowledge acquired in training from one or a limited number of views to other target views where recognition will be performed. They either transform view-dependent video features to a new view-invariant feature space [11,12], or adapt model parameters to the target views [3,4,26]. The transformation is learned on the co-occurrence statistics of view-dependent features.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Latent Multitask Learning for View-Invariant Action Recognition

Mahasseni

Todorovic

2013

2013 IEEE International Conference on Computer Vision

View full text Add to dashboard Cite

This paper presents an approach to view-invariant action recognition, where human poses and motions exhibit large variations across different camera viewpoints. When each viewpoint of a given set of action classes is specified as a learning task then multitask learning appears suitable for achieving view invariance in recognition. We extend the standard multitask learning to allow identifying: (1) latent groupings of action views (i.e., tasks), and (2) discriminative action parts, along with joint learning of all tasks. This is because it seems reasonable to expect that certain distinct views are more correlated than some others, and thus identifying correlated views could improve recognition. Also, part-based modeling is expected to improve robustness against self-occlusion when actors are imaged from different views. Results on the benchmark datasets show that we outperform standard multitask learning by 21.9%, and the state-of-the-art alternatives by 4.5-6%.

show abstract

Section: Latent Multitask Learningmentioning

confidence: 99%

Section: Latent Multitask Learningmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Latent Multitask Learning for View-Invariant Action Recognition

Mahasseni

Todorovic

2013

2013 IEEE International Conference on Computer Vision

View full text Add to dashboard Cite

show abstract

“…Our approach takes inspiration from recent works in cross-view data retrieval that tackle problems such as static cameras localization with satellite imagery [11], cross-view action recognition [16] and image-text retrieval [19,23]. These approaches achieve cross-view retrieval by learning the co-occurrence of features in different views.…”

Section: Introductionmentioning

confidence: 99%

Cross-View Image Geo-localization

Lin¹,

Belongie²,

Hays³

2016

Large-Scale Visual Geo-Localization

View full text Add to dashboard Cite

The recent availability of large amounts of geotagged imagery has inspired a number of data driven solutions to the image geolocalization problem. Existing approaches predict the location of a query image by matching it to a database of georeferenced photographs. While there are many geotagged images available on photo sharing and street view sites, most are clustered around landmarks and urban areas. The vast majority of the Earth's land area has no ground level reference photos available, which limits the applicability of all existing image geolocalization methods. On the other hand, there is no shortage of visual and geographic data that densely covers the Earthwe examine overhead imagery and land cover survey data -but the relationship between this data and ground level query photographs is complex. In this paper, we introduce a cross-view feature translation approach to greatly extend the reach of image geolocalization methods. We can often localize a query even if it has no corresponding groundlevel images in the database. A key idea is to learn the relationship between ground level appearance and overhead appearance and land cover attributes from sparsely available geotagged ground-level images. We perform experiments over a 1600 km 2 region containing a variety of scenes and land cover types. For each query, our algorithm produces a probability density over the region of interest.

show abstract

“…Typical examples can be found in [3], [8], [19], where only one action template is provided for each action class for training, and [15], where training samples are captured from a different viewpoint. In these situations, obtaining more labeled data is either impossible or expensive, while seeking for an alternative way of using data from other domains as compensation can be seen as a possible and economic solution.…”

Section: Introductionmentioning

confidence: 99%

Enhancing Action Recognition by Cross-Domain Dictionary Learning

Zhu¹,

Shao²

2013

Procedings of the British Machine Vision Conference 2013

View full text Add to dashboard Cite

We present a novel cross-dataset action recognition framework that utilizes relevant actions from other visual domains as auxiliary knowledge for enhancing the learning system in the target domain. The data distribution of relevant actions from a source dataset is adapted to match the data distribution of actions in the target dataset via a cross-domain discriminative dictionary learning method, through which a reconstructive, discriminative and domain-adaptive dictionary-pair can be learned. Using selected categories from the HMDB51 dataset as the source domain actions, the proposed framework achieves outstanding performance on the UCF YouTube dataset.

show abstract

Cross-view action recognition via view knowledge transfer

Cited by 248 publications

References 18 publications

Latent Multitask Learning for View-Invariant Action Recognition

Latent Multitask Learning for View-Invariant Action Recognition

Cross-View Image Geo-localization

Enhancing Action Recognition by Cross-Domain Dictionary Learning

Contact Info

Product

Resources

About