Bobick, A. & Wilson, A. (1997). A state-based approach to the representation and recognition of gesture. IEEE Transactions on Pattern Analysis and Machine Intelligence, 19(12), 1325--1337.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....progress in recent years at recognizing what people are doing. Most of the work in recognition of human activity has relied on recognizing a sequence of states using stochastic model based approaches. For example, hidden Markov models (HMMs) have become very popular for recognizing gestures [2, 12], sign language [17, 13] and actions [10, 18, 4] for a detailed review of these and other significant efforts in this direction, please review [1, 7] However, when it comes to recognizing activities with some predefined context or inherent semantics, purely probabilistic methods can be ....

A. F. Bobick and A. D. Wilson. A state based approach to the representation and recognition of gesture. PAMI, 19(12):1325--1337, December 1997.

....algorithms. They can perform dynamic time warping for structures that have been stretched and squashed in time. HMMs have been successfully applied to speech recognition [11] visual focus of attention [12] learning object movement and behaviour models [5, 8] and more recently gesture recognition [13, 2]. In the case of gesture recognition, the states are usually selected so as to capture the locations along the observation trajectories where measurements undergo significant change. Prior knowledge in the forms of state transition probabilities and conditional observation covariances are ....

A. Bobick and A. Wilson. A state-based approach to the representation and recognition of gesture. IEEE PAMI, 19(12):1325--1338, December 1997.

....as the temporal structure of relating temporally ordered visual events in space and time [19,21] Such temporal structures are often only considered to be first order for convenience. State transitions are learned from example sequences of visual events often manually clustered and labelled [20,31,4,6,21,15]. Methods for automatic temporal clustering of HMM states have also been proposed [5,30,54,55] Unfortunately, any process which operates on a representation has little, if any, effect on its semantic properties [56] However, a general assumption is often made such that the knowledge of a ....

A. Bobick and A.D. Wilson. A state-based approach to the representation and recognition of gesture. IEEE Trans. on Pattern Analysis and Machine Intelligence, 19(12):1325-1337, 1997.

....HMMs use dynamic programming to recognize the observation sequence. Recently, some state based approaches have been proposed for gesture modeling and recognition. The advantage of a state approach is that it doesn t need a large set of data in order to train the model. Bobick and Wilson [3] proposed an approach that models a gesture as a sequence of states in a configuration space. The training gesture data is first manually segmented and temporally aligned. A prototype curve is used to represent the data, and is parameterized according to a manually chosen arc length. Each segment ....

Aaron F. Bobick and Andrew D. Wilson. "A state-based approach to the representation and recognition of gesture," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 19, no. 12 Dec. 1997.

....temporal structures. A HMM can perform nonlinear dynamic time warping, stretching and squashing on temporal structures. HMMs have been applied to speech recognition [9] visual focus of attention [10] learning object movement and behaviour models [4, 7] and more recently gesture recognition [11, 2]. Hidden Markov states are usually selected so as to capture the locations along the observation trajectories where measurements undergo significant change. Prior knowledge in the form of state transition probabilities and conditional observation covariances are estimated from training examples. ....

A. Bobick and A. Wilson. A state-based approach to the representation and recognition of gesture. IEEE PAMI, 19(12):1325--1338, December 1997.

....and discussion (Section 5) 2. Related work Since the Moving Light Display experiments by Johansson [1] suggested that many human gestures could be recognized solely by motion information, motion profiles and trajectories have been investigated to recognize human gestures. Bobick and Wilson [2] proposed a state based technique for the representation and recognition of gesture. In their approach, a gesture is defined to be a sequence of states in a configuration space. The training gesture data is first reduced to a prototype curve through configuration space, and the prototype curve is ....

....sequence, as illustrated in Figure 2. By manually specifying the temporal sequence of states from the gesture examples, we obtain the structure of the Finite State Machine (FSM) for the gesture. For example, the state sequence for one cycle of the wave left hand gesture, shown in Figure 2, is [ 1 2 0 2 1 ]. Once this is determined, the training data is segmented into gesture samples. A sample of [ 1 1 1 2 2 2 2 0 0 0 0 2 2 2 1 1 ] for example, consists of the five states with (3, 4, 4, 3, and 2) samples per state, respectively. The number of samples in a state is proportional to the duration of ....

[Article contains additional citation context not shown here]

Aaron F. Bobick and Andrew D. Wilson. A State-Based Approach to the Representation and Recognition of Gesture. IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 19, no. 12 Dec. 1997.

....used a 3 D hand model to track a hand. They compared line features from the images with the projected model, and performed incremental state corrections. Similar work was presented by Kuch and Huang [5] in which the synthesis process could t the hand model to any person s hand. Bobick and Wilson [6] treated gesture as a sequence of states and computed con guration states along prototype gestures. Yacoob and Black proposed parameterized representation of human movement [7] Cutler and Davis [8] segmented the motion and computed a moving objects self similarity (including human motion ....

A. F. Bobick and A. D. Wilson. A state-based approach to the representation and recognition of gesture. IEEE Trans. on Pattern Analysis and Machine Intelligence, 19(12):1325-1337, December 1997.

....Similar work was presented by Kuch and Huang [7] where the synthesis process could fit the hand model to any person s hand. Tracking scenarios were used to verify its effectiveness. Delingette [8] used simplex meshes to recover the shape by connecting separately built models. Bobick and Wilson [9] treated gesture as a sequence of states and computed configuration states along prototype gestures. A review by Aggarwal and Cai [10] classified approaches to the human motion analysis, the tasks involved, and major areas related to human motion interpretation. Physically based modeling is ....

A.F. Bobick, A.D. Wilson, A state-based approach to the representation and recognition of gesture, IEEE Trans. on Pattern Analysis and Machine Intelligence 19 (12) (1997) 1325--1337.

....given the input from perceptual sensors and, if necessary, the previous state of the world. Kuniyoshi and Inoue [81] used finite automata to recognize human manipulations of blocks. Probabilistic methods for visual action recognition have been proposed in the computer vision community using HMMs [24, 26, 165] and stochastic grammars [22] However, as noted above, such approaches are not able to represent efficiently parallel threads of actions that occur in many everyday actions. Moreover, we believe that it is important to exploit the fact that logical impossibilities prevent the occurrence of some ....

....such as running and jogging. Gavrila and Davis [53] used 3D models to track upper body movements. Recent work by Bregler and Malik [29] proposes exponential maps and twist motions to simplify the recognition of movements of articulated structures such as the human body. Bobick and Wilson [24] proposed the use of sequences of states of eigen vector projection coefficients in a configuration space, allowing the direct use of imagery in the representation of movement. Davis and Bobick [43] proposed the use of view dependent temporal templates for the representation of full body movements ....

A. F. Bobick and A. D. Wilson. "A State-Based Approach to the Representation and Recognition of Gesture", IEEE PAMI, vol. 19 (12). 1997.

....a new observation is classified as being generated by the model that assigns the highest likelihood. Experiments on a set of 6 simple gestures, pick up, put down, push, pull, drop, and throw, demonstrate that gestures can be classified based on motion profiles. Bobick and Wilson [3] adopt a state based approach to represent and recognize gestures. First, many samples of a gesture are used to compute its principal curve [5] which is parameterized by arc length. A by product of calculating the curve is the mapping of each sample point of a gesture example to an arc length ....

....to the motions of different parts of, say, the palm, instead of a single representative point. Thus, each example of a gesture in our work is represented by a set of motion trajectories. Our experimental results show that an ensemble of trajectories yields better generalization Bobick and Wilson [3] adopt a state based approach to represent and recognize gestures. First, many samples of a gesture are used to compute its principal curve [5] which is parameterized by arc length. A byproduct of calculating the prototype is the mapping of each sample point of a gesture example to an arc length ....

[Article contains additional citation context not shown here]

A. F. Bobick and A. D. Wilson. A state-based approach to the representation and recognition of gesture. IEEE Trans. Pattern. Anal. Mach. Intell., 19(12):1325--1337, 1997.

....over the length of the gesture. Dynamic time warping (DTW) and Hidden Markov models (HMMs) are two techniques based on dynamic programming. Darrell and Pentland [12] applied DTW to match image template correlation scores against models to recognize hand gestures from video. In previous work [5], we represented gesture as a deterministic sequence of states through some configuration or feature space and employed a DTW parsing algorithm to recognize the gestures. The states were found by first determining a prototype gesture from a set of examples and then creating a set of states in ....

A.F. Bobick and A.D. Wilson, A State-Based Approach to the Representation and Recognition of Gesture,º IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 19, no. 12, pp. 1,325-1,337, Dec. 1997.

No context found.

Bobick, A. & Wilson, A. (1997). A state-based approach to the representation and recognition of gesture. IEEE Transactions on Pattern Analysis and Machine Intelligence, 19(12), 1325--1337.

No context found.

A.F. Bobick and A.D. Wilson. A state-based approach to the representation and recognition of gesture. PAMI, 19(12):1325--1337, Dec 1997.

No context found.

Aaron Bobick and Andrew Wilson. A State-based Approach to the Representation and Recognition of Gesture. IEEE Transactions on Pattern Analysis and Machine Intelligence, 19(12):1325--1337, 1997.

No context found.

A. F. Bobick and A. D. Wilson. A state-based approach to the representation and recognition of gesture. IEEE Trans. on PAMI, 19(12):1325--1337, December 1997.

No context found.

A. F. Bobick and A. D. Wilson. A state based approach to the representation and recognition of gesture. PAMI, 19(12):1325--1337, December 1997.

No context found.

A. F. Bobick and A. D. Wilson. A state based approach to the representation and recognition of gesture. PAMI, 19(12):1325--1337, December 1997.

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