TY - GEN
T1 - Robust incremental hidden conditional random fields for human action recognition
AU - Vrigkas, Michalis
AU - Mastora, Ermioni
AU - Nikou, Christophoros
AU - Kakadiaris, Ioannis A.
N1 - Publisher Copyright:
© Springer Nature Switzerland AG 2018.
PY - 2018
Y1 - 2018
N2 - Hidden conditional random fields (HCRFs) are a powerful supervised classification system, which is able to capture the intrinsic motion patterns of a human action. However, finding the optimal number of hidden states remains a severe limitation for this model. This paper addresses this limitation by proposing a new model, called robust incremental hidden conditional random field (RI-HCRF). A hidden Markov model (HMM) is created for each observation paired with an action label and its parameters are defined by the potentials of the original HCRF graph. Starting from an initial number of hidden states and increasing their number incrementally, the Viterbi path is computed for each HMM. The method seeks for a sequence of hidden states, where each variable participates in a maximum number of optimal paths. Thereby, variables with low participation in optimal paths are rejected. In addition, a robust mixture of Student’s t-distributions is imposed as a regularizer to the parameters of the model. The experimental results on human action recognition show that RI-HCRF successfully estimates the number of hidden states and outperforms all state-of-the-art models.
AB - Hidden conditional random fields (HCRFs) are a powerful supervised classification system, which is able to capture the intrinsic motion patterns of a human action. However, finding the optimal number of hidden states remains a severe limitation for this model. This paper addresses this limitation by proposing a new model, called robust incremental hidden conditional random field (RI-HCRF). A hidden Markov model (HMM) is created for each observation paired with an action label and its parameters are defined by the potentials of the original HCRF graph. Starting from an initial number of hidden states and increasing their number incrementally, the Viterbi path is computed for each HMM. The method seeks for a sequence of hidden states, where each variable participates in a maximum number of optimal paths. Thereby, variables with low participation in optimal paths are rejected. In addition, a robust mixture of Student’s t-distributions is imposed as a regularizer to the parameters of the model. The experimental results on human action recognition show that RI-HCRF successfully estimates the number of hidden states and outperforms all state-of-the-art models.
KW - Hidden markov model action recognition
KW - Student’s t-distribution hidden conditional random fields
UR - http://www.scopus.com/inward/record.url?scp=85057178709&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85057178709&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-03801-4_12
DO - 10.1007/978-3-030-03801-4_12
M3 - Conference contribution
AN - SCOPUS:85057178709
SN - 9783030038007
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 126
EP - 136
BT - Advances in Visual Computing - 13th International Symposium, ISVC 2018, Proceedings
A2 - Xu, Kai
A2 - Lin, Stephen
A2 - Boyle, Richard
A2 - Alsallakh, Bilal
A2 - Turek, Matt
A2 - Ramalingam, Srikumar
A2 - Bebis, George
A2 - Parvin, Bahram
A2 - Yang, Jing
A2 - Ventura, Jonathan
A2 - Koracin, Darko
A2 - Cuervo, Eduardo
PB - Springer-Verlag
T2 - 13th International Symposium on Visual Computing, ISVC 2018
Y2 - 19 November 2018 through 21 November 2018
ER -