Representation for Action Recognition Using Trajectory-Based Low-Level Local Feature and Mid-Level Motion Feature
The dense trajectories and low-level local features are widely used in action recognition recently. However, most of these methods ignore the motion part of action which is the key factor to distinguish the different human action. This paper proposes a new two-layer model of representation for actio...
Saved in:
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Wiley
2017-01-01
|
Series: | Applied Computational Intelligence and Soft Computing |
Online Access: | http://dx.doi.org/10.1155/2017/4019213 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
_version_ | 1832561790494441472 |
---|---|
author | Xiaoqiang Li Dan Wang Yin Zhang |
author_facet | Xiaoqiang Li Dan Wang Yin Zhang |
author_sort | Xiaoqiang Li |
collection | DOAJ |
description | The dense trajectories and low-level local features are widely used in action recognition recently. However, most of these methods ignore the motion part of action which is the key factor to distinguish the different human action. This paper proposes a new two-layer model of representation for action recognition by describing the video with low-level features and mid-level motion part model. Firstly, we encode the compensated flow (w-flow) trajectory-based local features with Fisher Vector (FV) to retain the low-level characteristic of motion. Then, the motion parts are extracted by clustering the similar trajectories with spatiotemporal distance between trajectories. Finally the representation for action video is the concatenation of low-level descriptors encoding vector and motion part encoding vector. It is used as input to the LibSVM for action recognition. The experiment results demonstrate the improvements on J-HMDB and YouTube datasets, which obtain 67.4% and 87.6%, respectively. |
format | Article |
id | doaj-art-b124b4c8393c45ba8a698e8c9f47ae62 |
institution | Kabale University |
issn | 1687-9724 1687-9732 |
language | English |
publishDate | 2017-01-01 |
publisher | Wiley |
record_format | Article |
series | Applied Computational Intelligence and Soft Computing |
spelling | doaj-art-b124b4c8393c45ba8a698e8c9f47ae622025-02-03T01:24:18ZengWileyApplied Computational Intelligence and Soft Computing1687-97241687-97322017-01-01201710.1155/2017/40192134019213Representation for Action Recognition Using Trajectory-Based Low-Level Local Feature and Mid-Level Motion FeatureXiaoqiang Li0Dan Wang1Yin Zhang2School of Computer Engineering and Sciences, Shanghai University, Shanghai 200444, ChinaSchool of Computer Engineering and Sciences, Shanghai University, Shanghai 200444, ChinaSchool of Computer Engineering and Sciences, Shanghai University, Shanghai 200444, ChinaThe dense trajectories and low-level local features are widely used in action recognition recently. However, most of these methods ignore the motion part of action which is the key factor to distinguish the different human action. This paper proposes a new two-layer model of representation for action recognition by describing the video with low-level features and mid-level motion part model. Firstly, we encode the compensated flow (w-flow) trajectory-based local features with Fisher Vector (FV) to retain the low-level characteristic of motion. Then, the motion parts are extracted by clustering the similar trajectories with spatiotemporal distance between trajectories. Finally the representation for action video is the concatenation of low-level descriptors encoding vector and motion part encoding vector. It is used as input to the LibSVM for action recognition. The experiment results demonstrate the improvements on J-HMDB and YouTube datasets, which obtain 67.4% and 87.6%, respectively.http://dx.doi.org/10.1155/2017/4019213 |
spellingShingle | Xiaoqiang Li Dan Wang Yin Zhang Representation for Action Recognition Using Trajectory-Based Low-Level Local Feature and Mid-Level Motion Feature Applied Computational Intelligence and Soft Computing |
title | Representation for Action Recognition Using Trajectory-Based Low-Level Local Feature and Mid-Level Motion Feature |
title_full | Representation for Action Recognition Using Trajectory-Based Low-Level Local Feature and Mid-Level Motion Feature |
title_fullStr | Representation for Action Recognition Using Trajectory-Based Low-Level Local Feature and Mid-Level Motion Feature |
title_full_unstemmed | Representation for Action Recognition Using Trajectory-Based Low-Level Local Feature and Mid-Level Motion Feature |
title_short | Representation for Action Recognition Using Trajectory-Based Low-Level Local Feature and Mid-Level Motion Feature |
title_sort | representation for action recognition using trajectory based low level local feature and mid level motion feature |
url | http://dx.doi.org/10.1155/2017/4019213 |
work_keys_str_mv | AT xiaoqiangli representationforactionrecognitionusingtrajectorybasedlowlevellocalfeatureandmidlevelmotionfeature AT danwang representationforactionrecognitionusingtrajectorybasedlowlevellocalfeatureandmidlevelmotionfeature AT yinzhang representationforactionrecognitionusingtrajectorybasedlowlevellocalfeatureandmidlevelmotionfeature |