Human action recognition with salient trajectories

Yi, Yang; Lin, Yikun

doi:10.1016/j.sigpro.2013.05.002

Cited by 33 publications

(13 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Other works have focused on trajectory-based approach, instead of extracting spatiotemporal interest points, they use trajectories to describe the video [19]. Yi et al [25] extracted salient trajectories by considering both their appearance and motion saliency. Similarly, Atmosukarto et al [4] extracted trajectories and used Fisher Kernel to create video representation.…”

Section: Related Workmentioning

confidence: 99%

Action Recognition Using Discriminative Structured Trajectory Groups

Atmosukarto

Ahuja

Ghanem

2015

2015 IEEE Winter Conference on Applications of Computer Vision

View full text Add to dashboard Cite

Section: Related Workmentioning

confidence: 99%

Action Recognition Using Discriminative Structured Trajectory Groups

Atmosukarto

Ahuja

Ghanem

2015

2015 IEEE Winter Conference on Applications of Computer Vision

View full text Add to dashboard Cite

“…It implies that although plenty of information can be obtained by dense descriptor, most of them are redundant even irrelevant. Therefore, Yi et al [30] aimed to deal with the problem of HAR with salient trajectories, then the salient trajectories were encoded by a hierarchical representation-based method. Iveel et al [29] extracted motion trajectories after video segment and build a bag-of-features (BOF) model based on four different types of descriptors (i.e., HOG, HOF, MBH, and TS).…”

Section: Related Workmentioning

confidence: 99%

“…Iveel et al [29] extracted motion trajectories after video segment and build a bag-of-features (BOF) model based on four different types of descriptors (i.e., HOG, HOF, MBH, and TS). Unlike [30,31] use video saliency to generate trajectory, Seo et al [16] introduced a trajectory rejection technology to avoid generating redundant trajectories and skip the frames that do not have much movement information, so the complexity of the algorithm is reduced.…”

Section: Related Workmentioning

confidence: 99%

Action recognition using length-variable edge trajectory and spatio-temporal motion skeleton descriptor

Weng

Guan

2018

J Image Video Proc.

View full text Add to dashboard Cite

Representing the features of different types of human action in unconstrained videos is a challenging task due to camera motion, cluttered background, and occlusions. This paper aims to obtain effective and compact action representation with length-variable edge trajectory (LV-ET) and spatio-temporal motion skeleton (STMS). First, in order to better describe the long-term motion information for action representation, a novel edge-based trajectory extracting strategy is introduced by tracking edge points from motion without limiting the length of trajectory; the end of the tracking is depending not only on the optical flow field but also on the optical flow vector position in the next frame. So, we only make use of a compact subset of action-related edge points in one frame to generate length-variable edge trajectories. Second, we observe that different types of action have their specific trajectory. A new trajectory descriptor named spatio-temporal motion skeleton is introduced; first, the LV-ET is encoded using both orientation and magnitude features and then the STMS is computed by motion clustering. Comparative experimental results with three unconstrained human action datasets demonstrate the effectiveness of our method.

show abstract

“…It has received great attention in the last two decades due to its wide applicability in many areas, such as surveillance, virtual reality, medical analysis, computer animation and human-computer interaction [2,3]. Although a large amount of research on human tracking has been carried out, the task of 3D human motion tracking from monocular image is still challenging due to following reasons.…”

Section: Introductionmentioning

confidence: 99%