Learned temporal models of image motion

Yacob, Yasmin Mohd; Davis, Larry S.

doi:10.1109/iccv.1998.710757

Cited by 10 publications

(5 citation statements)

References 10 publications

(9 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Event-based recognition attempts to segment image sequences by identifying events happening in a given context. One approach to simplify the task is to restrict events to a fixed set of predefined events [35,44]. Another is to restrict the shape of the objects and the context in which they appear [6,8].…”

Section: Previous Workmentioning

confidence: 99%

“…Recently, the system was extended to handle real image sequences in a restrictive set-up in which it recognized small set of actions [36]. Yacoob and Davis [44] describe an approach for image motion estimation that uses learned models of temporal-flows to recognize actions, such as various types of walking. Activities are learned from the temporal-flow models and represented as a set of orthogonal temporal-flow bases that are learned using principal component analysis of instantaneous flow measurements.…”

Section: Previous Workmentioning

confidence: 99%

“…Identifying general spatial motions is a hard problem, as discussed in the introduction. Identifying human motion is of particular interest [44]. In our framework, it requires first the development of a motion language for the domain (classical dance, soccer, exercising).…”

Section: Scope Extensionsmentioning

confidence: 99%

See 2 more Smart Citations

Understanding mechanical motion: From images to behaviors

Dar

Joskowicz

Rivlin

1999

Artificial Intelligence

View full text Add to dashboard Cite

We present an algorithm for producing behavior descriptions of planar fixed axes mechanical motions from image sequences using a formal behavior language. The language, which covers the most important class of mechanical motions, symbolically captures the qualitative aspects of objects that translate and rotate along axes that are fixed in space. The algorithm exploits the structure of these motions to robustly recover the objects behaviors. It starts by identifying the independently moving objects, their motion parameters, and their variation with respect to time using normal optical flow analysis, iterative motion segmentation, and motion parameter estimation. It then produces a formal description of their behavior by identifying individual uniform motion events and simultaneous motion changes, and parsing them with a motion grammar. We demonstrate the algorithm on three sets of image sequences: mechanisms, everyday situations, and a robot manipulation scenario.

show abstract

Section: Previous Workmentioning

confidence: 99%

Section: Previous Workmentioning

confidence: 99%

See 1 more Smart Citation

Understanding mechanical motion: From images to behaviors

Dar

Joskowicz

Rivlin

1999

Artificial Intelligence

View full text Add to dashboard Cite

show abstract

“…For representation, most use standard or generalized cylinders [1], [2], [3], but simpler planar [4] and more sophisticated quadrics-based [5], [6], [7] and deformable [8] models have been explored. The most widely used feature in matching between image and model is the edge, e.g., [5], [9], [10], [11], but increasing use is made of internal features, such as corners [12] and image motion [2], [3], [13], [14]. The key distinction in fitting pose to the image data is between works that adopt statistical techniques based on simple unimodal pdf's and solve deterministically, e.g., [2], [10], and those that represent arbitrary multimodal pdf's using mixture models or particle filters [3], [9], [15].…”

Section: Introductionmentioning

confidence: 99%

Recovering articulated pose: a comparison of two pre and postimposed constraint methods

Campos

Tordoff

Murray

2006

IEEE Trans. Pattern Anal. Machine Intell.

View full text Add to dashboard Cite

Abstract-We contrast the performance of two methods of imposing constraints during the tracking of articulated objects, the first method preimposing the kinematic constraints during tracking and, thus, using the minimum degrees of freedom, and the second imposing constraints after tracking and, hence, using the maximum. Despite their very different formulations, the methods recover the same pose change. Further comparisons are drawn in terms of computational speed and algorithmic simplicity and robustness, and it is the last area which is the most telling. The results suggest that using built-in constraints is well-suited to tracking individual articulated objects, whereas applying constraints afterward is most suited to problems involving contact and breakage between articulated (or rigid) objects, where the ability to test tracking performance quickly with constraints turned on or off is desirable.

show abstract

“…Optical flow is another popular approach to performing region based measurements [18,21,52,82,99,103,108], however it was again determined that the requirement of registered object templates was too restrictive.…”

Section: Discussionmentioning

confidence: 99%

Techniques for binocular markerless visual tracking of 3D articulated bodies

Smith¹

View full text Add to dashboard Cite

This thesis advances methods for performing markerless visual tracking of articulated bodies using one or two cameras. The research presented aims to improve upon existing Bayesian inspired tracking methods, by examining the 'building blocks' of these tracking algorithms, in particular the measurement function design, the state space selection, and local optimization methods. Results presented in this thesis show that improvements can be made in all of these areas. These improvements are applicable to a variety of Bayesian tracking algorithms.This thesis begins by examining literature relevant to the visual tracking problem. This includes the measurement functions used by other authors, focussing on the edge detection methods used in both tracking and segmentation problems. A general overview of the global search problem is given next, as a global search is a fundamental part of a Bayesian tracking algorithm. The combination of Newton like local optimization methods and the measurement functions used in visual tracking problems is examined next, and it is shown that Newton optimizers are not ideally suited to these measurement functions. The Bayesian tracking framework is then detailed, along with a review of several existing Bayesian tracking algorithms. Finally some non Bayesian tracking algorithms are discussed.Following the literature review, details of the models used in the experiments presented in this thesis are given. These include the articulated human body model, the camera model, image gradient metrics, self occlusion treatment, and a generic colour based region measurement method.The use of graph based approaches for edge measurements is then investigated. Graph based methods are commonly used in image segmentation problems, however have not been applied to visual tracking problems. A novel method for performing edge measurements using the 'shortest path' around the object's occluding contour is presented. Unlike in the segmentation problem, self occlusion models mean the weights or costs of some graph vertices can not be determined. Different treatments for occluded graph vertices are given and evaluated. It is shown that the graph based approach produces observational likelihoods that are more accurate and have significantly fewer local maxima than the edge measurement schemes previously used in tracking problems. While this approach is computationally more expensive than other methods, it is argued that this is offset by the reduced computational expense of the global search procedure used in tracking algorithms.The choice of state space used in the tracking problems is examined next. While most authors have used a state space based on the joint angles of the human body, a Cartesian state space based on the world coordinates of limbs is proposed. While Cartesian based state spaces have been used by iii other authors for representations of kinematic models, to the author's knowledge they have not been used for full kinematic models. It is shown that that the more linear relationship betwee...

show abstract

Learned temporal models of image motion

Cited by 10 publications

References 10 publications

Understanding mechanical motion: From images to behaviors

Understanding mechanical motion: From images to behaviors

Recovering articulated pose: a comparison of two pre and postimposed constraint methods

Techniques for binocular markerless visual tracking of 3D articulated bodies

Contact Info

Product

Resources

About