Exploring Motion Boundary based Sampling and Spatial-Temporal Context Descriptors for Action Recognition

@inproceedings{Peng2013ExploringMB,
  title={Exploring Motion Boundary based Sampling and Spatial-Temporal Context Descriptors for Action Recognition},
  author={Xiaojiang Peng and Yu Qiao and Qiang Peng and Xianbiao Qi},
  booktitle={BMVC},
  year={2013}
}
Motion Boundary : the maximum of gradient magnitudes between the horizontal and the vertical components of optical flow. • Implementation : 1. Sampling : sample points in current frame on a grid by a step size w at S spatial scales. Two kinds of points will be removed: (1) the minimal eigenvalue of the covariance matrix of derivatives is less than a given threshold T1. (2) in the background of the binary mask estimated from motion boundary. 2. Tracking and Filtering : the same with Wang’s [1]. 
Highly Cited
This paper has 46 citations. REVIEW CITATIONS

5 Figures & Tables

Topics

Statistics

0510201620172018
Citations per Year

Citation Velocity: 7

Averaging 7 citations per year over the last 3 years.

Learn more about how we calculate this metric in our FAQ.