Wrist Camera - Instruction Manual
Info
In the context of multiple objects detection, each object detected has its own set of detection score values (%).
Edges detection threshold and score
If the object is found, you will see the object outlined, surrounded by a blue rectangle, with the detection score value (%). If no object is
found, an error message will display, reading that the object was not found.
Other objects detected will be surrounded by a white rectangle. Tap the other object(s) to display their own set of detection score
values.
Fig. 4-49: Object found with detection score.
When testing the object(s) locating, object contours and detected features are outlined in green, while features that cannot be located
within the field of view of the Wrist Camera are outline in red. Features from the model that cannot be located on the object in the field
of view of the camera are outlined in red.
Info
In the context of multiple objects detection, each object detected has its own set of detection score values (%). When
performing the localization test, place the whole object over the workplane. Due to the perspective effect, some or all of the
object features might not be recognized. If an important feature needs to be found for your application (a hole for example),
make sure it is found during all localization tests. Again, green contour should match the object outlines at all times.
Tip
How to prevent false positives: To avoid false detections, remove the object from the workplane, decrease the detection
threshold to 0% and try to locate the object. If a false detection occurs on your workplane, you will see an object detected
with the detection score. Increase the detection threshold above this score to avoid false detection.
l
Try all areas of the workplane on which the object might be found. Adjust the detection threshold properly for the object and for the
runtime environment.
l
Adjust the detection threshold with the plus (+) and minus (-) buttons.
75
