Sort Poses by Input Scores¶
Function¶
This Step sorts poses by their scores and outputs the sorted list and its index list.
Input: the pose list to be sorted, the score list for the pose list. Please note that the two lists’ lengths should be the same. The scores usually come from the output port Pose Matching Scores of a 3D Fine Matching Step, or output ports for confidences from deep learning Steps.
Output: the sorted pose list, the index list of the sorted list.
Sample Scenario¶
Scenario 1¶
For machine tending tasks of metal workpieces, poses can be sorted based on the matching scores, and therefore workpieces with the most accurately calculated poses can be prioritized for picking.
Pose list to be sorted.
Matching score list corresponding to the pose list.
Sorted pose list.
Scenario 2¶
Sorting poses and associated masks based on the mask areas output from a deep learning Step.
Pose list to be sorted.
Mask list to be sorted.
Mask area list.
Sorted pose list.
Index list of the sorted pose list.
Reordered mask list based on the sorting result.
Parameters¶
Sort Order
Description: Decide whether to sort in ascending or descending order.Values:-Ascending
-Descending
Sample Data Flow¶
Figure 3 is an example of data flow in pose sorting based on matching scores in descending order.
Pose list to be sorted.
Matching score list corresponding to the pose list.
Sorted pose list.
Index list of the sorted pose list.
Comparison of Similar Steps¶
Sort Poses by Input Scores sorts in the same way as Sort 3D Poses when the parameter Sort Method of Sort 3D Poses is set to SORT_BY_CONFIDENCE, as shown in Figure 4.
Sort Poses by Input Scores outputs the index list of the sorted poses.
Sort Poses by Input Scores can also be used with a Reorder Step to reorder lists of data related to poses, such as masks.