Multiple Hypothesis Video Segmentation from Superpixel Flows

Vázquez-Reina, Amelio; Avidan, Shai; Pfister, Hanspeter; Miller, Eric L.

doi:10.1007/978-3-642-15555-0_20

Cited by 112 publications

(117 citation statements)

References 25 publications

(38 reference statements)

Supporting

Mentioning

111

Contrasting

Unclassified

Order By: Relevance

“…Recent works on video segmentation focus only on salient moving objects by analyzing point trajectories, while taking background as a single cluster [2], [29]. Some other works [3], [6] over-segment frames into superpixels, and partition them spatially and match them temporally. These methods provide a desirable computational reduction and powerful within-frame representation [30].…”

Section: Trends In Engineering and Technology (Nctet-2k17) Internatiomentioning

confidence: 99%

“…In order to create new labels or remove old labels when the objects enter or leave the camera view, we utilize a reasonable strategy to determine the label mapping by their spatial overlap [6]. An overlap of one frame between neighboring windows is used to determine whether current labels are new ones or mapped from previous ones.…”

Section: Unsupervised Video Segmentationmentioning

confidence: 99%

See 1 more Smart Citation

An Effective Approach to Streaming Video Segmentation with Sub-Optimal Low Rank Decomposition

Sujatha¹,

Babu²

2017

Nctet-2k17

View full text Add to dashboard Cite

Abstract-This paper explores how to perform powerful and proficient video division while smothering the impacts of information commotions and additionally debasements, and a viable approach is acquainted with this end. Initial, a general calculation, called problematic low-rank decay (SOLD), is proposed to seek after the low-rank portrayal for video division. Given the information framework shaped by supervoxel elements of a watched video succession, SOLD looks for a problematic arrangement by making the lattice rank unequivocally decided. Specifically, the portrayal coefficient grid with the settled rank can be disintegrated into two sub-networks of low rank, and after that we iteratively streamline them with shut frame arrangements. Additionally, we join a discriminative replication earlier into SOLD in light of the perception that little size video designs have a tendency to repeat regularly inside a similar question. Second, in light of SOLD, we introduce an effective deduction calculation to perform gushing video division in both unsupervised and intuitive situations. All the more particularly, the obliged standardized cut calculation is embraced by joining the low-rank portrayal with other low level signs and fleeting reliable imperatives for spatiotemporal division. Broad analyses on two open testing informational collections VSB100 and Seg Track propose that our approach beats other video division approaches in both exactness and effectiveness.

show abstract

Section: Trends In Engineering and Technology (Nctet-2k17) Internatiomentioning

confidence: 99%

Section: Unsupervised Video Segmentationmentioning

confidence: 99%

An Effective Approach to Streaming Video Segmentation with Sub-Optimal Low Rank Decomposition

Sujatha¹,

Babu²

2017

Nctet-2k17

View full text Add to dashboard Cite

show abstract

“…Some of the other methods make use of the superpixels. They connect them spatially and temporally to generate temporally consistent object regions [15,4,19,11]. This approach is often used in an unsupervised setting, which usually leads to severe over-segmentation.…”

Section: Related Workmentioning

confidence: 99%

Hierarchy of Localized Random Forests for Video Annotation

Nagaraja

Ochs

Liu

et al. 2012

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. We address the problem of annotating a video sequence with partial supervision. Given the pixel-wise annotations in the first frame, we aim to propagate these labels ideally throughout the whole video. While some labels can be propagated using optical flow, disocclusion and unreliable flow in some areas require additional cues. To this end, we propose to train localized classifiers on the annotated frame. In contrast to a global classifier, localized classifiers allow to distinguish colors that appear in both the foreground and the background but at very different locations. We design a multi-scale hierarchy of localized random forests, which collectively takes a decision. Cues from optical flow and the classifier are combined in a variational framework. The approach can deal with multiple objects in a video. We present qualitative and quantitative results on the Berkeley Motion Segmentation Dataset.

show abstract

“…For example, the key segments work [19] proposes a method to take frame-by-frame superpixel segmentations and automatically segment the dominant moving actor in the video with category independence. Recent works in video segmentation generate spatiotemporally coherent segmentations relatively efficiently by methods like point trajectory grouping [6,15,21], superpixel tracking [4,29,32], probabilistic methods [1,7,18], supervoxels by minimum spanning trees [16,33,34], or compositing multiple constituent segmentations [22,26]. These advances in video segmentation have also been thoroughly evaluated.…”

Section: Introductionmentioning

confidence: 99%

Flattening Supervoxel Hierarchies by the Uniform Entropy Slice

Whitt

Corso

2013

2013 IEEE International Conference on Computer Vision

View full text Add to dashboard Cite

Supervoxel hierarchies provide a rich multiscale decomposition of a given video suitable for subsequent processing in video analysis. The hierarchies are typically computed by an unsupervised process that is susceptible to undersegmentation at coarse levels and over-segmentation at fine levels, which make it a challenge to adopt the hierarchies for later use. In this paper, we propose the first method to overcome this limitation and flatten the hierarchy into a single segmentation. Our method, called the uniform entropy slice, seeks a selection of supervoxels that balances the relative level of information in the selected supervoxels based on some post hoc feature criterion such as objectness. For example, with this criterion, in regions nearby objects, our method prefers finer supervoxels to capture the local details, but in regions away from any objects we prefer coarser supervoxels. We formulate the uniform entropy slice as a binary quadratic program and implement four different feature criteria, both unsupervised and supervised, to drive the flattening. Although we apply it only to supervoxel hierarchies in this paper, our method is generally applicable to segmentation tree hierarchies. Our experiments demonstrate both strong qualitative performance and superior quantitative performance to state of the art baselines on benchmark internet videos.

show abstract

Multiple Hypothesis Video Segmentation from Superpixel Flows

Cited by 112 publications

References 25 publications

An Effective Approach to Streaming Video Segmentation with Sub-Optimal Low Rank Decomposition

An Effective Approach to Streaming Video Segmentation with Sub-Optimal Low Rank Decomposition

Hierarchy of Localized Random Forests for Video Annotation

Flattening Supervoxel Hierarchies by the Uniform Entropy Slice

Contact Info

Product

Resources

About