Generalization with Deep Learning: For Improvement on Sensing Capability by Zhenghua Chen Min Wu and Xiaoli Li

Author:Zhenghua Chen, Min Wu and Xiaoli Li
Language: eng
Format: epub
Publisher: World Scientific Publishing Co. Pte. Ltd.
Published: 2021-07-15T00:00:00+00:00

3.3Spatial connection configuration for skeleton

We follow the spatial connection configuration for a skeleton in Ref. [8] including three adjacent matrices. In general, for the ith joint ji in a skeleton, its neighbor can be defined as shown in Fig. 1(a). For instance, the neighbor of joint #1 is {#2, #3, #4). Leveraging the spatial configuration proposed in Ref. [38], the neighboring joints can be decomposed into three subsets, depending on the distance to the nose of the skeleton. More specifically, these three subsets can be denoted as follows: (i) the ith joint self; (ii) the neighboring joints closer to the nose than the ith joint; (iii) the neighboring joints farther away from the nose than the ith joint. After normalizing the whole adjacent matrix, we split it, namely, into three sub-matrices with the above criterion. Thus, Eq. (2) can be reformulated to

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.