Multiplicative kernels: Object detection, segmentation and pose estimation

Yuan, Quan; Thangali, Ashwin; Ablavsky, Vitaly; Sclaroff, Stan

doi:10.1109/cvpr.2008.4587739

Cited by 16 publications

(9 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…A closely related work to ours is Yuan et al [16], who learn "parameter sensitive detectors" for binary classification. Pose parameterization in [16] is achieved by multiplying a pose kernel K θ (θ, θ ) with the original kernel, and inference is performed by discretizing the pose and testing pose-specific classifiers.…”

Section: Related Workmentioning

confidence: 99%

“…(1) over the entire pose space, we can efficiently sample pose-specific detectors from the unified model. For any fixed θ, our model reduces to a single linear classifier w θ (this observation is also made by [16]):…”

Section: Pruning Stepmentioning

confidence: 99%

“…Pose parameterization in [16] is achieved by multiplying a pose kernel K θ (θ, θ ) with the original kernel, and inference is performed by discretizing the pose and testing pose-specific classifiers. While we also use a multiplicative decomposition of the joint kernel, we formulate the problem in the structured-output domain, and produce continuous solutions for pose during inference.…”

Section: Related Workmentioning

confidence: 99%

See 2 more Smart Citations

Parameterizing Object Detectors in the Continuous Pose Space

Sigal

Sclaroff

2014

Computer Vision – ECCV 2014

Self Cite

View full text Add to dashboard Cite

Abstract. Object detection and pose estimation are interdependent problems in computer vision. Many past works decouple these problems, either by discretizing the continuous pose and training pose-specific object detectors, or by building pose estimators on top of detector outputs. In this paper, we propose a structured kernel machine approach to treat object detection and pose estimation jointly in a mutually benificial way. In our formulation, a unified, continuously parameterized, discriminative appearance model is learned over the entire pose space. We propose a cascaded discrete-continuous algorithm for efficient inference, and give effective online constraint generation strategies for learning our model using structural SVMs. On three standard benchmarks, our method performs better than, or on par with, state-of-the-art methods in the combined task of object detection and pose estimation.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Pruning Stepmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Parameterizing Object Detectors in the Continuous Pose Space

Sigal

Sclaroff

2014

Computer Vision – ECCV 2014

Self Cite

View full text Add to dashboard Cite

show abstract

“…An alternative approach is to directly model the 3D viewpoint [16,19,18]. Recently, [15] showed that it is possible to learn a 3D part based representation that explicitly includes the viewpoint, which is also recovered as part of the detection process.…”

Section: Related Workmentioning

confidence: 99%

Pose estimation for category specific multiview object localization

Özuysal

Lepetit

Fua

2009

2009 IEEE Conference on Computer Vision and Pattern Recognition

135

View full text Add to dashboard Cite

show abstract

“…The features used to split training samples are learned between objects and non-objects, not between intra-class objects. In [23], Yuan et al proposed a multiplicative form of two kernel functions to learn the similarity for foreground-background and within-class variation jointly. The detectors associated with each foreground training sample are clustered to form the final set of detectors.…”

Section: Related Workmentioning

confidence: 99%

Robust multi-view car detection using unsupervised sub-categorization

Kuo

Nevatia

2009

2009 Workshop on Applications of Computer Vision (WACV)

View full text Add to dashboard Cite

This paper presents a novel approach for multi-view car detection using unsupervised sub-categorization instead of manual labeling. Cars have large variability of models and the view-point makes the appearance change dramatically. For object classes with a large intra-class variation like cars, a divide-and-conquer strategy may be applied. Instead of using manually predefined intra-class subcategorization, we examine several non-linear dimension reduction methods and group samples in the low-dimension embedding in an unsupervised way. The clustered samples have strong view-point similarities internally. A boostingbased cascade tree classifier is trained based on these subcategorizations. To demonstrate the capability of our multiview car detector, we create a more challenging test set with annotations. Compared to the UIUC side-view car data set, our test set contains a large range of car models, view points, and complex backgrounds. We compare our approach with previous methods and the result shows that ours outperforms the state-of-the-art methods.

show abstract

Multiplicative kernels: Object detection, segmentation and pose estimation

Cited by 16 publications

References 31 publications

Parameterizing Object Detectors in the Continuous Pose Space

Parameterizing Object Detectors in the Continuous Pose Space

Pose estimation for category specific multiview object localization

Robust multi-view car detection using unsupervised sub-categorization

Contact Info

Product

Resources

About