input csv 'frames' vs video frames

[akesner1]

I have been working on getting keypoint tracking from SLEAP into VAME. I was able to get VAME to happily read SLEAP output CSVs by sort of reformatting them to the structure of DLC. It wasn't too bad, just had to insert a blank row at the top of the CSV and another as the 3rd row, then delete the first and 3rd columns of the SLEAP CSV to match the overall structure of the DLC CSV.

After doing that, VAME loaded the data fine and I have been through the steps and no errors. So VAME is definitely doing stuff with the CSVs. But, here is my question. DLC CSVs all start at the 1st frame of the video, and if no animal is there (e.g. recording started before person put mouse in arena) it just puts blanks or zeros in those rows. SLEAP on the otherhand, seems to just start outputting data once the mouse is seen, so the first row of data in SLEAP could start at frame 150 or 300 or whatever, when the mouse is first found. So, does VAME use the number in that first column (i.e. frame number), or does it reindex the data from the first row of the CSV and relate that to first frame of the video? That is, if the first row of my CSV has the number 155 in the first column (i.e. that first column relates to frame 155 of the video), will VAME know this? Or will it, e.g. during video clips, think that that row is frame 1 of the video. I hope this makes sense. In other words, I am wondering, int he above scenario, if I need to make 154 "blank" rows in my CSV bewteen the headers and the first "frame" of data, so that VAME will match this with the video?

Thanks for the help.