Originally posted by **Toach-2021** November 1, 2023
Hello,
I have a very naive question, but how can I use Kalman filter for running inference with GUI?
When I run, I always get the error "Kalman filter requires target instance count".
I already prepared some predicted frames in the video.
I would appreciate any advice.
Thank you in advance.
Originally posted by **Toach-2021** November 2, 2023
I would like to try the Kalman filter to smooth out the tracking because inevitably some individuals would not be detected or their locations would be blurred...
About Target Instance Count, I'm sorry but I could not find that option...
Maybe I missed but I would appreciate if you could tell me how to set it.
Best,
Originally posted by **roomrys** November 2, 2023
Apologies, I was writing from memory, but we recently made some changes to these parameters. There used to be a "Cull to Target Instance Count" parameter and "Target Number of Instances Per Frame" parameter that has now been _replaced_ (in desired functionality) with "Max number of tracks" in #1447 - basically removing instances over a set limit earlier on in the tracking.
Although it looks like the variable names for these three parameters are different and although they have been removed from the GUI, the variables have not been refactored to reflect the preference for "Max number of Tracks".
https://github.com/talmolab/sleap/blob/0b2fdd9282dee40cd275bb040f91e575754bb048/sleap/config/pipeline_form.yaml#L379-L395
So, this is a GUI bug. However! Since there hasn't been changes to the variable names, using `sleap-track` through the CLI will work as intended.
Sorry about that,
Liezl
Discussed in https://github.com/talmolab/sleap/discussions/1581