We propose a feature for action recognition called Trajectory-Set (TS), on top of the improved Dense Trajectory (iDT). The TS feature encodes only trajectories around densely sampled interest points, without any appearance features. Experimental results on the UCF50 action dataset demonstrates that TS is comparable to state-of-the-arts, and outperforms iDT; the accuracy of 95.0%, compared to 91.7% by iDT.
Kenji MATSUI
Hiroshima University
Toru TAMAKI
Hiroshima University
Bisser RAYTCHEV
Hiroshima University
Kazufumi KANEDA
Hiroshima University
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Kenji MATSUI, Toru TAMAKI, Bisser RAYTCHEV, Kazufumi KANEDA, "Trajectory-Set Feature for Action Recognition" in IEICE TRANSACTIONS on Information,
vol. E100-D, no. 8, pp. 1922-1924, August 2017, doi: 10.1587/transinf.2017EDL8049.
Abstract: We propose a feature for action recognition called Trajectory-Set (TS), on top of the improved Dense Trajectory (iDT). The TS feature encodes only trajectories around densely sampled interest points, without any appearance features. Experimental results on the UCF50 action dataset demonstrates that TS is comparable to state-of-the-arts, and outperforms iDT; the accuracy of 95.0%, compared to 91.7% by iDT.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2017EDL8049/_p
Copy
@ARTICLE{e100-d_8_1922,
author={Kenji MATSUI, Toru TAMAKI, Bisser RAYTCHEV, Kazufumi KANEDA, },
journal={IEICE TRANSACTIONS on Information},
title={Trajectory-Set Feature for Action Recognition},
year={2017},
volume={E100-D},
number={8},
pages={1922-1924},
abstract={We propose a feature for action recognition called Trajectory-Set (TS), on top of the improved Dense Trajectory (iDT). The TS feature encodes only trajectories around densely sampled interest points, without any appearance features. Experimental results on the UCF50 action dataset demonstrates that TS is comparable to state-of-the-arts, and outperforms iDT; the accuracy of 95.0%, compared to 91.7% by iDT.},
keywords={},
doi={10.1587/transinf.2017EDL8049},
ISSN={1745-1361},
month={August},}
Copy
TY - JOUR
TI - Trajectory-Set Feature for Action Recognition
T2 - IEICE TRANSACTIONS on Information
SP - 1922
EP - 1924
AU - Kenji MATSUI
AU - Toru TAMAKI
AU - Bisser RAYTCHEV
AU - Kazufumi KANEDA
PY - 2017
DO - 10.1587/transinf.2017EDL8049
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E100-D
IS - 8
JA - IEICE TRANSACTIONS on Information
Y1 - August 2017
AB - We propose a feature for action recognition called Trajectory-Set (TS), on top of the improved Dense Trajectory (iDT). The TS feature encodes only trajectories around densely sampled interest points, without any appearance features. Experimental results on the UCF50 action dataset demonstrates that TS is comparable to state-of-the-arts, and outperforms iDT; the accuracy of 95.0%, compared to 91.7% by iDT.
ER -