|
The following example shows the effect of weights on DTW matching on the match results. We vary the weights for different body areas. For example, we can give more importance to the upper-body joints and less to the lower-body joints. This is done by giving the upper-body joints larger weights, and lower-body joints smaller spatial thresholds. Additionally, by using negative weights we can lower the ranking of potential matches with similar motions to the query.
The query is a short happy running motion 96 keyframes long (or 4 seconds).
Matching parameters: DTW, rotational parameterization, full-body area, no global position, no global rotation.
DTW Example 1
|
Video Description
(in order of appearance) |
|
Note: wupper=0.5 means that the existing weights of the upper body degrees of freedom are multiplied by 0.5. These include the head, back, shoulders and arms. wlower refers to all remaining joints.
Effect of weights on DTW matching - 2 |
Same as above except that the query is a short sitting motion 56 keyframes long (or 2.3 seconds).
Matching parameters: DTW, rotational parameterization, full-body area, no global position, no global rotation.
DTW Example 2
|
Video Description
(in order of appearance) |
|
Note: wupper=0.5 means that the existing weights of the upper body degrees of freedom are multiplied by 0.5. These include the head, back, shoulders and arms. wlower refers to all remaining joints.