Analysing the streams from LSL_Relay
Hi, i used Pupil Player and uploaded my Invisible data. I am trying to understand what certain exported data mean - however i did not find an explanation for the following things: Using eye overlay, i got numpy arrays like these "eye0_lookup.npy", it contains (when i printed it in python) 4 values per row e.g. eye 0:
[(0, 0, 0.58940152, 0) (0, 1, 0.59736952, 717) (0, 2, 0.60145152, 1085) (0, 3, 0.60540752, 1441) (0, 4, 0.60935852, 1797) (0, 5, 0.61735652, 2516) (0, 6, 0.62140252, 2880) (0, 7, 0.62541252, 3244) (0, 8, 0.62970652, 3627) (0, 9, 0.63756352, 4335)] What does each of these values mean?
These data files are only meant to be used internally by Player. You'll generally want to use the export function to generate data in a more friendly format (CSV)
Is there a specific task you're trying to accomplish? Maybe we can help point you in the right direction
I understand but when i export, i only get excels/csv files containing timestamp data for the eyes. So my aim is to manually detect any closing of the eye. When my participants close their eyes for a longer time - it's not tracked as "blink" since it's technically not one. But i nevertheless would like to be able to manually calculate and track every time the eye is being closed (and the duration of that). I was hoping i get some data on the eye camera by using Pupil Player
The file you're looking at is likely just frame indices and timestamps. You're going to want data from the blink detector itself.
Looking at the code, blink data isn't saved to disk separately. It's just based on pupil detection confidence
But is this available for Pupil Invisible recordings in Pupil Player?
Hi @user-f0b6e1 , Pupil Invisible's blink detector uses a different algorithm than what is used for Pupil Core. The Pupil Invisible algorithm is run by Pupil Cloud, after the recordings are uploaded.
If you want to compute blinks locally, have a look at the pl-rec-export command line tool.
It seems like it's not based on the documentation for pupil player
On another note, i've been working with the fixations.csv (raw data derived from cloud by downloading the timeseries data, which contains all fixations) and the one i derive by downlaoding my AOI enrichment. I noticed that some fixations - naturally cannot be placed within the reference image these are flagged as "false" and do not contain any coordinates in the fixations.csv (from the AOI). This is perfectly fine. Some fixations however are missing entirely - e.g. missing in fixations.csv (AOI) but not from raw data. This is confusing, since while they were (correctly) not mapped onto the reference image, they are not the only fixations outside the image. Some others can be mapped and accurately flagged as "false" nevertheless with empty columns in coordinates. Why is it that way? Is this only the case for fixations that cannot be mapped on the reference image or also for valid ones? And what is the difference between fixations that are flagged as flase and ones that are completly missing from the csv file?
@user-f0b6e1 , would you be able to open a Support Ticket in 🛟 troubleshooting about this? Thanks.