Primitive Based Action Representation and recognition

Sanmohan Baby

Publikation: Bog/antologi/afhandling/rapportPh.d.-afhandlingForskning

974 Downloads (Pure)

Resumé

The presented work is aimed at designing a system that will model and
recognize actions and its interaction with objects. Such a system is aimed
at facilitating robot task learning. Activity modeling and recognition is
very important for its potential applications in surveillance, human-machine
interface, entertainment, biomechanics etc. Recent developments in neuroscience
suggest that all actions are a compositions of smaller units called
primitives.
Current works based on primitives for action recognition uses a supervised
framework for specifying the primitives. We propose a method to
extract primitives automatically. These primitives are to be used to generate
actions based on certain rules for combining. These rules are expressed
as a stochastic context free grammar. A model merging approach is adopted
to learn a Hidden Markov Model to t the observed data sequences. The
states of the HMM approximates local properties of a long sequence.
Observation sequences are used to learn a model expressing the data in
a structured way. Based on the learned model, recurring parts in the sequences
are identied. Primitives that make up the observation sequences
are identied as the recurring and unique parts appearing in the sequences.
Extracted primitives are used to make a primitive graph from which a grammar
for the observed primitives are derived.
This method is further extended to include object context. We observe
that human actions and objects can be seen as being intertwined: we can
interpret actions from the way the body parts are moving, but as well from
how their eect on the involved object. While human movements can look
vastly dierent even under minor changes in location, orientation and scale,
the use of the object can provide a strong invariant for the detection of
motion primitives. Movements that produce the same state change in the
object state space are classied to be instances of the same action primitive.
This allows us to dene action primitives as sets of movements where the
movements of each primitive are connected through the object state change
they induce.
OriginalsprogDansk
ForlagComputer Vision and Media Technology Laboratory (CVMT), Aalborg University
Antal sider105
StatusUdgivet - 2010

Citer dette

Baby, S. (2010). Primitive Based Action Representation and recognition. Computer Vision and Media Technology Laboratory (CVMT), Aalborg University.
Baby, Sanmohan. / Primitive Based Action Representation and recognition. Computer Vision and Media Technology Laboratory (CVMT), Aalborg University, 2010. 105 s.
@phdthesis{29c746d81da7482b942493a4ef431183,
title = "Primitive Based Action Representation and recognition",
abstract = "The presented work is aimed at designing a system that will model and recognize actions and its interaction with objects. Such a system is aimed at facilitating robot task learning. Activity modeling and recognition is very important for its potential applications in surveillance, human-machine interface, entertainment, biomechanics etc. Recent developments in neuroscience suggest that all actions are a compositions of smaller units called primitives. Current works based on primitives for action recognition uses a supervised framework for specifying the primitives. We propose a method to extract primitives automatically. These primitives are to be used to generate actions based on certain rules for combining. These rules are expressed as a stochastic context free grammar. A model merging approach is adopted to learn a Hidden Markov Model to t the observed data sequences. The states of the HMM approximates local properties of a long sequence. Observation sequences are used to learn a model expressing the data in a structured way. Based on the learned model, recurring parts in the sequences are identied. Primitives that make up the observation sequences are identied as the recurring and unique parts appearing in the sequences. Extracted primitives are used to make a primitive graph from which a grammar for the observed primitives are derived. This method is further extended to include object context. We observe that human actions and objects can be seen as being intertwined: we can interpret actions from the way the body parts are moving, but as well from how their eect on the involved object. While human movements can look vastly dierent even under minor changes in location, orientation and scale, the use of the object can provide a strong invariant for the detection of motion primitives. Movements that produce the same state change in the object state space are classied to be instances of the same action primitive. This allows us to dene action primitives as sets of movements where the movements of each primitive are connected through the object state change they induce.",
author = "Sanmohan Baby",
year = "2010",
language = "Dansk",
publisher = "Computer Vision and Media Technology Laboratory (CVMT), Aalborg University",

}

Baby, S 2010, Primitive Based Action Representation and recognition. Computer Vision and Media Technology Laboratory (CVMT), Aalborg University.

Primitive Based Action Representation and recognition. / Baby, Sanmohan.

Computer Vision and Media Technology Laboratory (CVMT), Aalborg University, 2010. 105 s.

Publikation: Bog/antologi/afhandling/rapportPh.d.-afhandlingForskning

TY - BOOK

T1 - Primitive Based Action Representation and recognition

AU - Baby, Sanmohan

PY - 2010

Y1 - 2010

N2 - The presented work is aimed at designing a system that will model and recognize actions and its interaction with objects. Such a system is aimed at facilitating robot task learning. Activity modeling and recognition is very important for its potential applications in surveillance, human-machine interface, entertainment, biomechanics etc. Recent developments in neuroscience suggest that all actions are a compositions of smaller units called primitives. Current works based on primitives for action recognition uses a supervised framework for specifying the primitives. We propose a method to extract primitives automatically. These primitives are to be used to generate actions based on certain rules for combining. These rules are expressed as a stochastic context free grammar. A model merging approach is adopted to learn a Hidden Markov Model to t the observed data sequences. The states of the HMM approximates local properties of a long sequence. Observation sequences are used to learn a model expressing the data in a structured way. Based on the learned model, recurring parts in the sequences are identied. Primitives that make up the observation sequences are identied as the recurring and unique parts appearing in the sequences. Extracted primitives are used to make a primitive graph from which a grammar for the observed primitives are derived. This method is further extended to include object context. We observe that human actions and objects can be seen as being intertwined: we can interpret actions from the way the body parts are moving, but as well from how their eect on the involved object. While human movements can look vastly dierent even under minor changes in location, orientation and scale, the use of the object can provide a strong invariant for the detection of motion primitives. Movements that produce the same state change in the object state space are classied to be instances of the same action primitive. This allows us to dene action primitives as sets of movements where the movements of each primitive are connected through the object state change they induce.

AB - The presented work is aimed at designing a system that will model and recognize actions and its interaction with objects. Such a system is aimed at facilitating robot task learning. Activity modeling and recognition is very important for its potential applications in surveillance, human-machine interface, entertainment, biomechanics etc. Recent developments in neuroscience suggest that all actions are a compositions of smaller units called primitives. Current works based on primitives for action recognition uses a supervised framework for specifying the primitives. We propose a method to extract primitives automatically. These primitives are to be used to generate actions based on certain rules for combining. These rules are expressed as a stochastic context free grammar. A model merging approach is adopted to learn a Hidden Markov Model to t the observed data sequences. The states of the HMM approximates local properties of a long sequence. Observation sequences are used to learn a model expressing the data in a structured way. Based on the learned model, recurring parts in the sequences are identied. Primitives that make up the observation sequences are identied as the recurring and unique parts appearing in the sequences. Extracted primitives are used to make a primitive graph from which a grammar for the observed primitives are derived. This method is further extended to include object context. We observe that human actions and objects can be seen as being intertwined: we can interpret actions from the way the body parts are moving, but as well from how their eect on the involved object. While human movements can look vastly dierent even under minor changes in location, orientation and scale, the use of the object can provide a strong invariant for the detection of motion primitives. Movements that produce the same state change in the object state space are classied to be instances of the same action primitive. This allows us to dene action primitives as sets of movements where the movements of each primitive are connected through the object state change they induce.

M3 - Ph.d.-afhandling

BT - Primitive Based Action Representation and recognition

PB - Computer Vision and Media Technology Laboratory (CVMT), Aalborg University

ER -

Baby S. Primitive Based Action Representation and recognition. Computer Vision and Media Technology Laboratory (CVMT), Aalborg University, 2010. 105 s.