Graph-based transductive learning for robust visual tracking

Zha, Yufei; Yang, Yuan; Du-yan, BI

doi:10.1016/j.patcog.2009.06.011

Cited by 34 publications

(20 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We measure the quality of our proposal in terms of precision and recall comparing the system with the results in the similar work of Coppi et al [1] where only the positive labeled model was exploited. We also compare our method with the baseline method of Zha et al [16], that recently proposed the use of graph-based transductive learning for visual tracking without any update strategy. Tab.…”

Section: Methodsmentioning

confidence: 99%

“…1 and 2 contain respectively average recall and precision results obtained from THIS, PETS2004, PETS2009, and CampusVideo using our Transductive Tracer with Positive and Negative examples, TTPN, performed both with and without the proposed update strategy and compares these values to the results obtained with the tracing system proposed in [1] and with the baseline method in [16]. [16] We point out how the presented method outperforms the baseline algorithm and method with only the positive model. Both precision and recall are significantly improved by the negative model exploitation instead of the single positive model.…”

Section: Methodsmentioning

confidence: 99%

“…In the recent years SSL has been used in many areas including computer vision where several applications i.e. image segmentation [4] or object recognition and tracking [16] are solved with a semisupervised classification. The reason for this choice can be explained observing that in many real world applications it is relatively easy to acquire a large amount of unlabeled (unclassified) data whereas it is quite difficult and expensive to have labeled, or classified, data.…”

Section: Introductionmentioning

confidence: 99%

“…The tracing engine exploits semi supervised learning (SSL) to detect among the candidates the positive samples of the target objects, if present. Differently from previously proposed methods [16,1], this transductive learning approach based on Riemannian space uses both positive and negative examples to classify new elements. Since the target appearance can evolve in time, we designed the update engine to retain the best target images and capture most of its appearance variability.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

People appearance tracing in video by spectral graph transduction

Coppi

Calderara

Cucchiara

2011

2011 IEEE International Conference on Computer Vision Workshops (ICCV Workshops)

View full text Add to dashboard Cite

Following people in different video sources is a challenging task: variations in the type of camera, in the lighting conditions, in the scene settings (e.g. crowd or occlusions) and in the point of view must be accounted. In this paper we propose a system based only on appearance information that, disregarding temporal and spatial information, can be flexibly applied on both moving and static cameras. We exploit the joint use of transductive learning and spectral properties of graph Laplacians proposing a formulation of the people tracing problem as a semi-supervised classification. The knowledge encoded in two labeled input sets of positive and negative samples of the target person and the continuous spectral update of these models allow us to obtain a robust approach for people tracing in surveillance video sequences. Experiments on publicly available datasets show satisfactory results and exhibit a good robustness in dealing with short and long term occlusions.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

People appearance tracing in video by spectral graph transduction

Coppi

Calderara

Cucchiara

2011

2011 IEEE International Conference on Computer Vision Workshops (ICCV Workshops)

View full text Add to dashboard Cite

show abstract

“…We show retrospective learning is important for correcting errors in the past. Transductive approaches [26,27] are limited by the fact that the general transductive problem is highly non-convex and hard to solve. We show that transduction can be effectively applied for the isolated subproblem of frame selection (for self-paced learning).…”

Section: Related Workmentioning

confidence: 99%

Self-Paced Learning for Long-Term Tracking

Supancic

Ramanan

2013

2013 IEEE Conference on Computer Vision and Pattern Recognition

212

146

View full text Add to dashboard Cite

We address the problem of long-term object tracking, where the object may become occluded or leave-the-view. In this setting, we show that an accurate appearance model is considerably more effective than a strong motion model. We develop simple but effective algorithms that alternate between tracking and learning a good appearance model given a track. We show that it is crucial to learn from the "right" frames, and use the formalism of self-paced curriculum learning to automatically select such frames. We leverage techniques from object detection for learning accurate appearance-based templates, demonstrating the importance of using a large negative training set (typically not used for tracking). We describe both an offline algorithm (that processes frames in batch) and a linear-time online (i.e. causal) algorithm that approaches real-time performance. Our models significantly outperform prior art, reducing the average error on benchmark videos by a factor of 4.

show abstract