Training does not work with multiple videos!

[transkriptase]

Hello,

i try to create a custom made good model to run more tracking on high computing system of my uni and/or on GPU computer, but I could not manage to do it even I try the basic commands. Is there a reason why? Because I know people who is working with multiple videos from differen species of same animal (mine are same species and same setup). the command I run is simple sleap-train baseline.centrod.json (I also tried bottomup) --video_paths "video1" "video2"......video25"

And the error I get is
"TypeError: '>' not supported between instances of 'NoneType' and 'int'"

Can you please help me!
Best,

[eberrigan]

Hi @transkriptase,

Please see our CLI reference here: https://sleap.ai/develop/guides/cli.html.

Can you confirm SLEAP is installed first with sleap-train --help.

• I am not sure where your files are located relative to your current directory, but getting all of the video-paths correct can be a pain. You can export a sleap-labels package in the GUI on your desktop computer and then train using that on your HPC.
• Please try a comma delimiter between video-paths CLI definition
• --video-paths is only used when the videos in your labels file are not accessible.
• Is your labels_path specified in your training config file?

Please copy and paste your exact command and error to help troubleshoot the error.

Thanks,

Elizabeth