Hi,
I am gap-filling motion capture trials using Vicon Nexus. Typically, this is a very straightforward process. However, one of our test subjects is very difficult to gap fill because the marker label quality seems poor. For example, the markers more frequently become unlabeled or swapped throughout the trial. I noticed that the number of trajectories for this participant is extremely high (~2,000). Does this mean there is reflective material being picked up and causing a 'labeling' confusion? Are there ways to reduce the number of trajectories to a much smaller number so that the number of trajectories is closer to the number of desired testing markers in the capture volume? I appropriately masked the cameras and calibrated them before collecting data.
Thanks,
David Hollinger
Here is a picture of the capture volume. Note the number of trajectories in the bottom left corner.
Hi David, that’s a common problem for me as well. You may try and right click on a ghost trajectory and choose to completely delete it from your recordings. It may also help to not auto-label but manually label the first frame while selecting the forwards mode (between the forward/backward options), and move frame by frame doing adjustments and deletions until the end of the recording.
Thanks for the recommendations. I followed @Sokratis Dimitrios Komaris tip of manual labeling the first frame rather than auto-label. However, the cause of the extra trajectories came from a reflective material on the edge of our capture volume. During our next data collection, we covered it with a thin fabric sheet, which seemed to solve the issue. Our calibration image error was < 0.2 for all cameras and our marker placements usually are spread out by a few inches at least
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Programming Languages
- Python