Learning to Segment and Track in RGBD

Teichman, Alex; Lussier, Jake T.; Thrun, Sebastian

doi:10.1007/978-3-642-36279-8_35

Cited by 22 publications

(17 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The goal of our algorithm is to collect training examples of objects in unstructured environments from RGBD sequences. Our method is different from (Teichman et al, 2013) in multiple, fundamental aspects. First of all, the authors introduce the use of a large number of features instead of only geometric cues computed to avoid redundant computational complexity.…”

Section: Related Workmentioning

confidence: 94%

“…They also provided multiple baseline algorithms under two main categories: depth as an additional cue and point cloud tracking. Among the ten proposed variations the The most similar, recent work to ours is that of (Teichman et al, 2013), which uses online learning to segment consecutive RGBD frames. However, this work assumes a single initial segmentation is provided by human labeling.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Non-Rigid Multi-Body Tracking in RGBD Streams

Dai

Guo

Mordohai

et al. 2019

ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

<p><strong>Abstract.</strong> To efficiently collect training data for an off-the-shelf object detector, we consider the problem of segmenting and tracking non-rigid objects from RGBD sequences by introducing the spatio-temporal matrix with very few assumptions &ndash; no prior object model and no stationary sensor. Spatial temporal matrix is able to encode not only spatial associations between multiple objects, but also component-level spatio temporal associations that allow the correction of falsely segmented objects in the presence of various types of interaction among multiple objects. Extensive experiments over complex human/animal body motions with occlusions and body part motions demonstrate that our approach substantially improves tracking robustness and segmentation accuracy.</p>

show abstract

Section: Related Workmentioning

confidence: 94%

Section: Related Workmentioning

confidence: 99%

Non-Rigid Multi-Body Tracking in RGBD Streams

Dai

Guo

Mordohai

et al. 2019

ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

show abstract

“…The individual steps shown in Figure 5: player selection/background subtraction, skeleton tracking and gestures were obtained from the Omek Beckon™ SDK 3.0 Windows Edition-Developer's Guide, included in the installation of the SDK. There are several methodologies and techniques in the literature to perform background subtracting based on ToF cameras [54,55] and based on structured light cameras [56,57]. …”

Section: System Designmentioning

confidence: 99%

Use of a Time-of-Flight Camera With an Omek Beckon™ Framework to Analyze, Evaluate and Correct in Real Time the Verticality of Multiple Sclerosis Patients during Exercise

Eguíluz

García

2013

IJERPH

View full text Add to dashboard Cite

Any person with Multiple Sclerosis (MS), regardless of the severity of their disability, needs regular physical activity. Poorly performed exercises could aggravate muscle imbalances and worsen the patient’s health. In this paper, we propose a human body verticality detection system using a time-of-flight camera as a tool to detect incorrect postures and improve them in real time. The prototype uses Omek’s Beckon™ Framework to analyze and evaluate the position of patients during exercise. Preliminary results, based on objective questionnaires, indicate an improvement in patients’ evolution through better positions and performance of the exercises.

show abstract

“…We evaluate our method using the Stanford Track Collection [20], a dataset of about 13,000 tracks (a total of about 1.3 million frames) extracted from natural, unstaged suburban environments with a dense LIDAR system mounted on a car. Data was recorded while driving and while parked at busy intersections with many people, bicyclists, and cars.…”

Section: A Datasetmentioning

confidence: 99%

Tracking-based semi-supervised learning

Teichman

Thrun

2012

The International Journal of Robotics Research

Self Cite

View full text Add to dashboard Cite

Abstract-In this paper, we consider a semi-supervised approach to the problem of track classification in dense 3D range data. This problem involves the classification of objects that have been segmented and tracked without the use of a class model.We propose a method based on the EM algorithm: iteratively 1) train a classifier, and 2) extract useful training examples from unlabeled data by exploiting tracking information. We evaluate our method on a large multiclass problem in dense LIDAR data collected from natural suburban street scenes. When given only three hand-labeled training tracks of each object class, semi-supervised performance is comparable to that of the fully-supervised equivalent which uses thousands of hand-labeled training tracks. Further, when given additional unlabeled data, the semi-supervised method outperforms the supervised method.Finally, we show that a simple algorithmic speedup based on incrementally updating a boosting classifier can reduce learning time by a factor of three.

show abstract

Learning to Segment and Track in RGBD

Cited by 22 publications

References 23 publications

Non-Rigid Multi-Body Tracking in RGBD Streams

Non-Rigid Multi-Body Tracking in RGBD Streams

Use of a Time-of-Flight Camera With an Omek Beckon™ Framework to Analyze, Evaluate and Correct in Real Time the Verticality of Multiple Sclerosis Patients during Exercise

Tracking-based semi-supervised learning

Contact Info

Product

Resources

About