Mixed markers and video on Qualisys system

[opherdonchin]

We want to use one or two cameras in our Qualisys system to record videos for later markerless tracking of some objects. This seems like a common application, but the manual seems confusing on how best to proceed. Let's say we need 10 seconds of recording (repeated 300 times), is there a way to work backwards from that to get the appropriate settings for the video comrpession?

The application is that we are tracking a person shooting a shot in billiards with markers and the balls themselves we plan to track with markerless tracking using DeepLabCut or similar. If there is a better way to approach the problem, I'd be happy to know that as well.

Thanks,
Opher

[Edmund Cramp]

Opher, your data collection efforts reminds me of the early data collection testing efforts that I first saw in early gait lab setups in the 80's, hopefully if you contact your data collection system manufacturer, or the people who installed the system for you, then they will help you setup the environment to collect your data. I have always seen these data collection testing efforts as universally helpful to all 3D data collection uses, so posting more information about the testing and the results here will be helpful for everyone.

[opherdonchin]

Thanks. I'm hacking a lab set up for gait measurement to measure billiard shots. We're using Arcus and Opus cameras for the marker tracking and the video. We plan to use 1 camera for video and the other cameras (14?) for tracking. We really don't need more than 6 seconds of recording at once.

We have access to support from both the national representative and the manufacturer, but those things can take a while. That's no slight on either of them who are available and ready to help, but it takes a while to get everyone on the page. All I want to get some preliminary data collected to see whether the entire experiment.

I think we've figured it out, but if anyone has good references on or experience with the varied options in the Qualisys system for compression in the video, as well as insight into how best to optimize the tradeoff between comrpession and recording length, that would be interesting to hear about.

Thanks,
Opher

[JohSverker]

Good to hear that you have figured out a solution, if you have any further questions feel free to contact us directly on support@qualisys.com. We can help you understand the available options that you have on your Oqus and Arqus models.

Regards,
Johannes Sverker
Support Team Lead
Qualisys AB

[opherdonchin]

Hi Johannes,

Thanks for dropping in on this! As long as you're here, maybe you could help out with my question.

How does a user of your system figure out what is determining their maximum number of frames for video collection from one of your cameras? Is there a formula? How does the different compression options or other parameters play into this?

The "best practice" values in your manual are very helpful, but it's not at all clear how to change things if I need to get an extra 10 frames or double the number of frames I need.

Thanks,
Opher

[JohSverker]

Hi Opher,

If you only have the Arqus cameras in your system, then they save the raw video data on the internal memory. So if you want to capture more video frames then you have to change the Image size setting to reduce the number of pixels or use the Sensor mode option to reduce the resolution and keep the FOV.

You also mentioned having Oqus cameras. According to our logs these might be the 5+ model. These can be set to In-camera MJPEG on the Video compression option. The video frames are then streamed during the capture and you are not limited by the internal memory. You can find the maximum streaming rates for the different sensor modes in the chapter "Qualisys video sensor specifications (in-camera MJPEG)" in Appendix A in the manual. You can also combine the Sensor mode with a reduced Image size and get a higher frame rate.

However the best option for video capture in QTM in is the Miqus video cameras, which gives you a color image and higher streaming rates, https://www.qualisys.com/cameras/miqus-video/.

Regards,
Johannes

[opherdonchin]

Thanks Johannes, that's very helpful.

Do I understand correctly from your answer that the compression choices have no effect on the either the filling of the internal buffer in the case of the arcus or the streaming rate in the case of the Oqus? The only two parameters to be concerned with are the image size and the sensor mode?

Opher

[JohSverker]

Hi Opher,

Yes you are correct. For Arqus the raw video is saved in the camera. The compression setting only changes the size and quality of the video file on the computer.

If you use the In-camera MJPEG option then the quality option changes the quality of the video, but since the video is streamed during the measurement it doesn't change the streaming rate that you can use. This is because the limitation is in the MJPEG compression and not in the data transfer.

Regards,
Johannes