N. Oliver, B. Rosario, and A. Pentland, "A Bayesian computer vision system for modeling human interactions," IEEE Trans. Pattern Anal. Machine Intell., vol. 22, pp. 831--843, Aug. 2000.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... the segmentation and tracking problems have been somehow solved, and the focus is on the issues related to temporal segmentation [15, 22] Another type looks at scenes in which people are small blobs where their articulated structure is unimportant when seen from cameras mounted high on buildings [21]. Yet another type of approach looks at articulated structures of people, but simplifies the detection of silhouettes by placing people in front of featureless background [23] With the proposed approach, we can analyze the video streams produced by fixed surveillance cameras. We can retrieve ....

N. Oliver, B. Rosario, A. Pentland, A Bayesian Computer Vision System for Modeling Human Interactions, IEEE Trans. PAMI, vol. 22 (8), pp. 831--843, 2000.

.... Alternatively, Hidden Markov Models (HMMs) 19] have been widely used for tackling simple behaviours such as gestures or gait recognition [21, 20] Other extensions to the basic HMM have also been used such as the Coupled Hidden Markov Models (CHMMs) for modeling human behaviours and interactions [15], and variable length Markov models (VLMMs) to locally optimise the size of behaviour models [8] All these approaches employ a flat model of activities. To develop scalable systems for high level behaviour recognition, we need a framework that utilizes the inherent hierarchical structure. ....

N. M. Oliver, B. Rosario, and A. Pentland. A Bayesian computer vision system for modeling human interactions. IEEE Transactions on Pattern Analysis and Machine Intelligence, 22(8):831--843, 2000.

....this approach requires a great deal of training data. This is also true of parameterised HMMs [23] which can su er from lack of stability in the interpretation compared to deformable model tracking and analysis seen in the next section. More recent work by Oliver, Rosario and Pentland [34] [35] has developed reliable Bayesian vision systems. Two exciting recent developments are: 1) work by Galata, Johnson and Hogg using deformable models with HMM behaviour models for virtual actors [24] and 2) the action reaction learning of Jebara and Pentland [36] 37] which models interactions and ....

N. Oliver, B. Rosario, and A. Pentland, \A Bayesian computer vision system for modeling human interactions," in International Conference on Vision Systems, Gran Canaria, Spain, 1999.

....developing an event detection program that scans the log files for common events that can be given a semantic label. There has been much work on parsing sequences of low level surveillance observations, particularly time stamped trajectories, into events that signify object interactions [5] [10] [12] 15] 17] 87] 90] Our prototype event detector is based on hidden Markov models (HMMs) 10] 87] trained to recognize simple object interactions. Briefly, output from the low level detection and tracking surveillance algorithms is quantized into the following discrete set of ....

....a semantic label. There has been much work on parsing sequences of low level surveillance observations, particularly time stamped trajectories, into events that signify object interactions [5] 10] 12] 15] 17] 87] 90] Our prototype event detector is based on hidden Markov models (HMMs) [10], 87] trained to recognize simple object interactions. Briefly, output from the low level detection and tracking surveillance algorithms is quantized into the following discrete set of attributes and values for each motion blob: 1) object class: Human, Vehicle, HumanGroup; 2) object action: ....

[Article contains additional citation context not shown here]

N. M. Oliver, B. Rosario, and A. P. Pentland, "A Bayesian computer vision system for modeling human interactions," IEEE Trans. Pattern Anal. Mach. Intell., vol. 22, pp. 831--843, Aug. 2000.

....moving regions in a scene. This showed invariance to lighting variations but involved an costly, off line initialization. It primary application is for geometrically static backgrounds. Recently, an eigenvector approximation of the entire image was used to model the background in outdoor scenes[20]. Changes in scene lighting can cause problems for many backgrounding methods. Ridder et al. 21] mod eled each pixel with a Kalman Filter which made their system more robust to lighting changes in the scene. While this method does have a pixel wise automatic threshold, it still recovers slowly ....

Oliver, Nuria, Barbara Rosario, Alex Pentland. "A Bayesian Computer Vision System for Modeling Human Interactions, " Proceedings of ICVS99 Gran Canaria, Spain, January 1999.

....2. 1 Overview There is growing interest in computer vision and multimedia signal processing for understanding the behaviour of interacting people, for actions that are de ned by playing both similar and complementary roles (e.g. a handshake, a dancing couple, or a children s game) 5] 6] [7], 8] While most of the work for recognition of interactions has been directed towards visual surveillance in outdoor [7] and oce scenarios [6] the analysis of people interaction constitutes a richer research domain. Group interaction recognition can be approached probabilistically with models ....

.... of interacting people, for actions that are de ned by playing both similar and complementary roles (e.g. a handshake, a dancing couple, or a children s game) 5] 6] 7] 8] While most of the work for recognition of interactions has been directed towards visual surveillance in outdoor [7] and oce scenarios [6] the analysis of people interaction constitutes a richer research domain. Group interaction recognition can be approached probabilistically with models that handle multiple information streams and capture consistent data relationships. Within this framework, interaction ....

N. Oliver, B. Rosario, and A. Pentland. A bayesian computer vision system for modeling human interactions. IEEE Transactions on Pattern Analysis and Machine Intelligence, 22(8), August 2000.

.... video sequences, is done using recursive and neural networks, deformable templates, spatio temporal templates [60] and graphical models [12] because they offer dynamic time warping and a clear Bayesian semantics for both individual (HMM) and interacting or coupled (CHMM) generative processes [61]. Finally, some authors have implemented systems that combine detection, tracking, and recognition [2] 15] 20] 37] 39] 57] 58] 68] A second set of criteria that can be used for classifying research on human motions is based on how to model humans. Humans have been modeled as ....

N. M. Oliver, B. Rosario, and A. P. Pentland, "A Bayesian computer vision system for modeling human interactions," IEEE Trans. Pattern Anal. Machine Intell., vol. 22, pp. 831--843, 2000.

.... Principal Component Analysis (PCA) in particular is a popular technique for parameterizing shape, appearance, and motion [8, 4, 18, 19, 29] These learned PCA representations have proven useful for solving problems such as face and object recognition, tracking, detection, and background modeling [2, 8, 18, 19, 20]. Typically, the training data for PCA is pre processed in some way (e.g. faces are aligned [18] or is generated by some other vision algorithm (e.g. optical flow is computed from training data [4] As automated learning methods are applied to more realistic problems, and the amount of training ....

....people often pass though the view of the camera quickly, they sometimes remain relatively still over multiple frames. We applied standard PCA and RPCA to the training data to build a background model that captures the illumination variation. Such a model is useful for person detection and tracking [20]. The second column of Fig. 8 shows the result of reconstructing each of the illustrated training images using the PCA basis (with 20 basis vectors) The presence of people in the scene effects the recovered illumination of the background and results in ghostly images where the people are poorly ....

N. Oliver, B. Rosario, and A. Pentland. A Bayesian computer vision system for modeling human interactions. ICVS. Gran Canaria, Spain, Jan. 1999.

....methods to statistical outliers. In particular, PCA is a popular technique for parameterizing shape, appearance, and motion [8, 15, 48, 50, 67] Learned PCA representations have proven useful for solving problems such as face and object recognition, tracking, detection, and background modeling [5, 15, 48, 50, 52]. Typically, the training data for PCA is pre processed in some way (e.g. faces are aligned [48] or is generated by some other vision algorithm (e.g. optical flow is computed from training data [8] As automated learning methods are applied to more realistic problems, and the amount of training ....

....much lower values of ##:## and ##:##. 5 Experiments To illustrate the range of applications of Robust Subspace Learning (RSL) we consider two problems of current interest in computer vision. The first involves learning a background appearance model for use in person detection and tracking [52]. More generally, RSL can be applied to any other eigen image learning problem. We also consider the problem of computing structure from the motion of tracked feature points. We show how both of these problems benefit from a robust formulation that can reject intra sample outliers. 5.1 Learning a ....

[Article contains additional citation context not shown here]

N. Oliver, B. Rosario, and A. Pentland. A Bayesian computer vision system for modeling human interactions. In H. I. Christensen, editor, Int. Conf. Computer on Vision Systems, ICVS, volume 1542 of LNCS-Series, pages 255--272, Gran Canaria, Spain, January 1999. Springer-Verlag. 38

.... a method of classifying more detailed human movement patterns using a view based temporal template approach[Davis and Bobick, 1997] Brand et al. use coupled HMMs to represent gestures that involve both arms[Brand et al. 1997] Oliver et al. use the same framework to model human interactions[Oliver et al. 2000]. 3 Memory Assistant Overview The basic idea for the Memory Assistant is to have a set of pan, tilt, zoom cameras monitor specific locations in the home, such as the kitchen counter, bedroom dressers, etc. for movement of objects and agents, using a wide field of view. After local activity has ....

Nuria M. Oliver, Barbara Rosario, and Alex P. Pentland, "A Bayesian Computer Vision System for Modeling Human Interactions," IEEE Trans. on Pattern Analysis and Machine Intelligence, 22(8), August 2000.

....Hidden Markov Model rolled out in time. Circles represent the C HMM at di erent time steps. Squares represent observations. Figure 4: A Coupled Hidden Markov Model rolled out in time. Circles represent the C HMM at di erent time steps. Squares represent observations. Coupled Hidden Markov Models [1, 19] (CHMM ) attempts to model two O(N) processes without modeling the complete joint probability of O(N ) states and O(N ) observations. CHMM s, model the joint state distribution, but treat the observations as independent. The motivation for this construction is to increase the speed of ....

....observations and beliefs are limited and that they depend on each other. Clearly limiting the dependencies among beliefs 15 is essential in order to make the problem tractable. However, unlike Uncertainty and Utility, a Bayesian solution can take on many forms from a variety of Markov Models [1, 8, 19] to Dynamic Bayes Nets [5] and Dynamic Object Oriented Bayes Nets [7] In other work, Horvitz spells out more speci cs Figure 7: Calculating the optimal action [10] of Mixed Initiative Computing [10] He presents the following model of deciding whether or not an action should be taken: eu(A) ....

N. M. Oliver, B. Rosario, and A. Pentland. A Bayesian Computer Vision System for Modeling Human Interactions. IEEE Transactions on Pattern Analysis and Machine Intelligence, 22(8):831-843, 2000.

....scanpath sequences can be analysed for their similarity. They use a Markov model for the sequences and inspect the sequences with the help of string editing methods. Markov models were used in a human behavior model for surveillance systems that is controlled by top down and bottom up attention [18]. Rimey and Brown [19] have used a leftto right Augmented HMM to do classification based on the scanpath information. Another important study that makes use of the scanpath information is Rao et al. 20] where bottom up and top down influences are considered to plan saccades. In a recent study, ....

N. Oliver, B. Rosario, and A. Pentland, " Bayesian Computer Vision System for Modeling Human Interactions,"IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 2, no. 8, pp. 831-843, Aug. 2000.

....is intractable and needs approximation techniques. However an exact solution exists for 2 interacting chains that influence each other through causal links. Such a structure called Coupled HMMs [Brand96] has been effectively used in modeling simple human behavior between 2 interacting pedestrians [Oliver99]. The literature on learning a model of human behavior over time spans greater than a few seconds is sparse. Hogg et al. demonstrate how to learn characteristic motions of pedestrians in a plaza, by representing maps of pedestrian trajectories as non parametric distributions over several hours ....

Oliver, N., Rosario, B., Pentland, A. 1999. A Bayesian Computer Vision System for Modeling Human Interactions. Proceedings of ICVS'99, Gran Canaria, Spain.

....system, a specific interaction may be isolated from a long image sequence obtained by a surveillance video camera. Our system consists of two parts: 1) the human segmentation and tracking, and (2) the interaction classification. Much work has been done in the area of human segmentation [1 4,15]. In [4] Pentland et al. segmented humans by subtracting an eigenbackground, which was generated using Principal Component Analysis (PCA) from several static background images. They tracked humans based on their position and velocity, estimated using a Kalman filter and the Probability Density ....

....a specific interaction may be isolated from a long image sequence obtained by a surveillance video camera. Our system consists of two parts: 1) the human segmentation and tracking, and (2) the interaction classification. Much work has been done in the area of human segmentation [1 4,15] In [4], Pentland et al. segmented humans by subtracting an eigenbackground, which was generated using Principal Component Analysis (PCA) from several static background images. They tracked humans based on their position and velocity, estimated using a Kalman filter and the Probability Density Function ....

[Article contains additional citation context not shown here]

Nuria Oliver, Barbara Rosario and Alex Pentland, A Bayesian computer vision system for modeling human interactions , Proceedings of International Conference on Vision Systems '99, Gran Canaria, Spain, pp. 255-272, January 1999.

....collected. Principle component analysis is used to determine means and variances over the entire sequence (whole images as vectors) Incoming images are projected onto the PCA subspace. Differences between the projection and the current image greater than a threshold are considered foreground [9]. Linear Prediction [A] This is the algorithm described in Section 2.1. Wallflower [A] This is a combination of the pixel, region and frame level algorithms as described in Section 2. All of the test sequences were taken with a 3 CCD camera recording to digital tape. We stored the sequences ....

Oliver, N., B. Rosario, and A. Pentland. "A Bayesian Computer Vision System for Modeling Human Interactions." in Int'l Conf. on Vision Systems. 1999. Gran Canaria, Spain: Springer.

....do not fall into the background class. The fundamental assumption of the algorithm is that the background is static, or at most slowly varying, in all respects: geometry, reflectance, and illumination. These algorithms show low sensitivity to slow lighting changes, to which they need to adapt. (Oliver et al. 1999) uses a subtraction method which has an explicit illumination model. The model in this case is the eigenspace, describing a range of appearances of the scene under a variety of lighting conditions. The method works extremely well for outdoor environments with static geometry, but unforeseen ....

Oliver, N., B. Rosario, and A. Pentland: 1999, `A Bayesian Computer Vision System for Modeling Human Interactions'. In: Proceedings of ICVS99. Gran Canaria, Spain.

....enabling the recovery of the targets 3D trajectory. It is also shown that mutual occlusion occurs in a well defined configuration and can therefore be dealt with. 1 Introduction Surveillance applications are extremely varied and many of them have been subject of a significant research effort [1, 2, 3, 4, 5]. Besides the applications themselves, the techniques employed include a wide range of computer vision domains. For example highlevel modeling and reasoning [3, 4] is extremely important for the development of the applications since interpretation of the events is crucial for the usefulness of ....

....Surveillance applications are extremely varied and many of them have been subject of a significant research effort [1, 2, 3, 4, 5] Besides the applications themselves, the techniques employed include a wide range of computer vision domains. For example highlevel modeling and reasoning [3, 4] is extremely important for the development of the applications since interpretation of the events is crucial for the usefulness of automated surveillance. Motion segmentation is another issue with high relevance for surveillance applications. There is a wide range of approaches for this problem ....

N. Oliver, B. Rosario, and A. Pentland. A bayesian computer vision system for modeling human interactions. In ICVS99--First Int. Conf. on Computer Vision Systems, pages 255--272, 1999.

....provide images of the humans in specific areas. 1 Introduction Automated monitoring of human activity is important for many applications. The problem of analyzing human activity in video has been the focus of several researchers effort and several systems have been described in the literature [1, 2, 3, 4, 5, 6, 7, 8, 9]. Many of these systems consist of a computer vision system to detect and segment a moving object and a higher level interpretation module. In very specialized applications other sensors are used besides vision. Automatic interpretation of the data is very difficult and most systems in use ....

N. Oliver, B. Rosario, and A. Pentland. A bayesian computer vision system for modeling human interactions. In ICVS99--First Int. Conf. on Computer Vision Systems, pages 255--272, 1999.

No context found.

N. Oliver, B. Rosario, and A. Pentland, "A Bayesian computer vision system for modeling human interactions," IEEE Trans. Pattern Anal. Machine Intell., vol. 22, pp. 831--843, Aug. 2000.

No context found.

N. M. Oliver, B. Rosario, and A. P. Pentland, A Bayesian Computer Vision System for modeling human interactions, IEEE Trans. on Pattern Analysis and Machine Intelligence, vol. 22, 2000, pp. 831-843.

No context found.

N. Oliver, B. Rosario, and A. Pentland. A Bayesian computer vision system for modeling human interactions. ICVS. Gran Canaria, Spain, Jan. 1999.

No context found.

N. M. Oliver, B. Rosario, and A. P. Pentland, A Bayesian Computer Vision System for modeling human interactions, IEEE Trans. on Pattern Analysis and Machine Intelligence, vol. 22, 2000, pp. 831-843.

No context found.

N.M. Oliver, B. Rosario, and A.P. Pentland. A bayesian computer vision system for modeling human interactions. PAMI, 22(8):831--843, August 2000.

No context found.

N. Oliver, B. Rosario, and A. Pentland. A bayesian computer vision system for modeling human interactions. In Proceedings of ICVS99, Gran Canaria, Spain, 1999.

No context found.

Oliver, N., Rosario, B., Pentland, A., A Bayesian Computer Vision System for Modeling Human Interactions, IEEE Transactions on Pattern Analysis and Machine Intelligence, p.p. 831-843, August, 2000.

No context found.

N.M. Oliver, B. Rosario, and A.P. Pentland, "A Bayesian Computer Vision System for Modeling Human Interactions," IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 22, no. 8, pp. 831-843, Aug. 2000.

No context found.

N. Oliver, B. Rosario, and A. Pentland, "A Bayesian computer vision system for modeling human interactions," IEEE Trans. Pattern Anal. Mach. Intell., vol. 22, no. 8, pp. 831--843, Aug. 2000.

No context found.

N. Oliver, B. Rosario, and A. Pentland, "A bayesian computer vision system for modeling human interactions," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 22, August 2000.

No context found.

N. Oliver, B. Rosario, and A. Pentland, "A bayesian computer vision system for modeling human interactions," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 22, August 2000.

No context found.

N. Oliver, B. Rosario, and A. Pentland. A Bayesian computer vision system for modeling human interactions. In International Conference on Vision Systems, January 1999.

No context found.

N. Oliver, B. Rosario, and A. Pentland. A Bayesian computer vision system for modeling human interactions. IEEE Transactions on Pattern Analysis and Machine Intelligence, 22(8), August 2000.

No context found.

N. Oliver, B. Rosario, and A. Pentland. A Bayesian computer vision system for modeling human interactions. IEEE PAMI, 22(8), Aug. 2000.

No context found.

N. Oliver, B. Rosario, and A. Pentland. A Bayesian computer vision system for modeling human interactions. IEEE Transactions on Pattern Analysis and Machine Intelligence, 22(8), August 2000.

No context found.

N. Oliver, B. Rosario, and A. Pentland. A Bayesian computer vision system for modeling human interactions. ICVS. Gran Canaria, Spain, Jan. 1999.

No context found.

N. Oliver, B. Rosario and A. Pentland, "A Bayesian Computer Vision System for Modeling Human Interactions, " Proceedings of Intl. Conference on Vision Systems ICVS99. Gran Canaria. Spain. , January 1999.

No context found.

N. Oliver, B. Rosario, and A. Pentland. A bayesian computer vision system for modeling human interactions. In Proceedings of ICVS99, Gran Canaria, Spain, 1999.

No context found.

N. Oliver, B. Rosario, and A. Pentland. A bayesian computer vision system for modeling human interactions. IEEE Trans. on Pattern Analysis and Machine Intell., 22(8):831--843, 2000.

No context found.

Nuria Oliver, Barbara Rosario, and Alex Pentland. A bayesian computer vision system for modeling human interactions. In Proceedings of ICVS99,Gran Canaria, Spain, January 1999.

No context found.

N. Oliver, B. Rosario, and A. Pentland, "A bayesian computer vision system for modeling human interactions," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 22, August 2000.

No context found.

N.M. Oliver, B. Rosario, and A.P. Pentland. A bayesian computer vision system for modeling human interactions. PAMI, 22(8):831--843, August 2000.

No context found.

N. Oliver, B. Rosario, and A. Pentland, "A Bayesian computer vision system for modeling human interactions," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 22, no. 8, pp. 831--841, 2000.

No context found.

Oliver, N., Rosario, B., Pentland, A. A Bayesian Computer Vision System for Modeling Human Interactions. IEEE Transactions On Pattern Analysis And Machine Intelligence. Vol. 22, No. 8, August 2000.

No context found.

N. Oliver, B. Rosario, and A. Pentland, "A bayesian computer vision system for modeling human interactions," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 22, August 2000.

No context found.

N. Oliver, B. Rosario, and A. Pentland. A bayesian computer vision system for modeling human interactions. IEEE Transactions on Pattern Analysis and Machine Intelligence, 22(8), August 2000.

No context found.

N.M. Oliver, B. Rosario, and A.P. Pentland. A bayesian computer vision system for modeling human interactions. PAMI, 22(8):831--843, August 2000.

No context found.

Nuria M. Oliver, Barbara Rosario, Alex P. Pentland, "A Bayesian Computer Vision System for Modeling Human Interactions", PAMI August 2000 (Vol. 22, No. 8).

No context found.

N. M. Oliver, B. Rosario, A. P. Pentland, "A Bayesian Computer Vision System for Modeling Human Interactions," IEEE Transactions of Pattern Analysis and Machine Intelligence , Vol. 22, pp. 831-843, 2000.

No context found.

N. Oliver, B. Rosario, and A. Pentland, `A Bayesian Computer Vision System for Modeling Human Interactions', IEEE Trans. on Pattern Analysis and Machine Intelligence, 22(8), 831--843, (2000).

No context found.

Nuria M. Oliver, Barbara Rosario, and Alex P. Pentland, A Bayesian Computer Vision System for Modeling Human Interactions, IEEE Trans. on PAMI, vol. 22, no. 8, August 2000.

No context found.

N. Oliver, B. Rosario, and A. Pentland, \A bayesian computer vision system for modeling human interactions," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 22, August 2000.

First 50 documents  Next 50

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC