Citation Context

...f the surfaces. Performing inference over the combinatorial range of possible occlusion relationships is challenging and, consequently, only a few layered flow models explicitly encode relative depth =-=[12, 30]-=-. Recent work revisits the layered model to handle occlusions [9], but does not explicitly model the layer ordering or achieve state-of-the-art performance on the Middlebury benchmark. While most curr...

Citation Context

...atial information with temporal occlusions. Jojic et al. [6] and Tao et al. [14] both model videos as a layered composition of objects and use EM to infer objects appearances and motions. Zhou et al. =-=[19]-=- extend [14] by introducing the concept of background occluding layers and explicitly inferring depth ordering of foreground layers. Most of the above mentioned approaches rely on partial observations...

Citation Context

...luder-free region U t can be deduced from the incomplete visual hull VH t at time t. (b) U: occluder-free regions accumulated over time. tions for detecting one or several motion and occlusion layers =-=[4, 18]-=-. Such layers can be used and accounted for, when building the visual hull of dynamic objects [11]. To the best of our knowledge no method has addressed the dense recovery of 3D occluder shapes from m...

Citation Context

...oss track fragments corresponding to the same object. Recent work attempts to address these tracking issues by linking track fragments together in a multi-object framework [9] or by background layers =-=[18]-=-. We propose a novel algorithm that solves event recognition and track linking jointly. Rather than assuming complete tracks for event actors, we dynamically determine the global track linking solutio...

Citation Context

... pp. 337–346, 2005. c○ Springer-Verlag Berlin Heidelberg 2005s338 K. Kim, D. Harwood, and L.S. Davis Research on layers for motion segmentation, object tracking, or occlusion analysis can be found in =-=[20,21,22,23,24,25,26]-=-. [20] worked on motion and appearance in layers, [21] on subspace approach, [22] on Bayesian approach, [23] on depth ordering by tracking edges, [24] on transparency manifolds, [25] on depth layers f...

Citation Context

...affine degrees of freedom. In contrast by making a (fairly weak) assumption of temporal continuity, we can afford to represent alignments and their associated uncertainties analytically. Zhou and Tao =-=[13]-=- describe an approach to modelling the background which may occlude foreground layers. This work is similar to ours in formulation but does not consider multiple views and in some respects may be rega...

Citation Context

...f the moving object. This is particularly likely to happen when the background has large regions with almost constant color or intensity level. Layered models [53], [49], [8], [51], [34], [33], [21], =-=[55]-=- brought new approaches to the segmentation of moving objects. Tao et al. proposed a filtering approach where a two-dimensional (2-D) Gaussian shape model is updated from frame to frame [51]. This wor...

Citation Context

...d leave frequently, but the occlusion map must be available immediately for pedestrian tracking. Tracking through occlusions via a set of foreground and background layers was proposed by Zhou and Tao =-=[11]-=-. However, that framework did not model motion of objects around occluding layers. Our sampling-based inference algorithm for tracking is influenced by the work of Smith, et al. [7]. Their formulation...