Open aremazeilles opened 4 years ago
There is already a format for events but it is oriented to gait (gaitEvents.yaml). Other partners are considering to use a event (.csv) file with two columns: the timestamp of the event and the label of the event. I like gaitEvents.yaml format because it avoids repeating the same label when you store information about a repetitive movement (walking). So to make it more generic I would rename it to events.yaml and say that for the events related to gait, the proposed labels are: r_heel_strike, l_heel_strike, r_toe_off, l_toe_off.
The only limit I can imagine to the yaml mode is the frequency of events in the file. If it is quite small, then yaml sounds great. If it is higher, then the yaml format could have its limit, no?
Having a common format for gait events and other sounds great to me. Then the question is whether we should go for yaml, csv, or permit both.
I see advantages in higher frequencies for yaml (smaller size).
I am in favour of allowing only one. However the PiManager does not enter in how the content is formatted at this momment so both formats could be accepted. The name before the extension should be equal (events.csv/events.yaml)
I see advantages in higher frequencies for yaml (smaller size). High frequency (1000Hz) means a lot of value?
events.csv (99bytes): | time | event |
---|---|---|
0.001 | r_heel_strike | |
0.002 | l_heel_strike | |
0.003 | r_heel_strike | |
0.004 | l_heel_strike |
events.yaml (60bytes)
r_heel_strike: [0.001, 0.003]
l_heel_strike: [0.002, 0.004]
If the experiments last several seconds, then the yaml file will just be unreadbale, while the csv could be easier to read (even through automatic editor plugins although.
even through automatic editor plugins although.
Not sure about it. If we are talking about reading, both are comparable. If we need filtering/editing or other capabilities then csv plugins are better.
I see advantages in .csv if multiple events are allowed at same timestamp. If this feature is needed, then definitely we should go with .csv
An example od long event file in yaml format
Some \n
are to be removed, but this is more or less wuold be the yaml aspect with a significant number of items in it.
So if we stick to yaml, shall we change the scope of gaitEvents
to be events
, or do we add another format for event in general, maintaining still the one for gaitEvent ?
Note that if we go for first option, we would not have issues with people willing to mix gait info and other event labeled.
So I assume we would add here another format, for TimeStamped event, defined as a yaml file.
I reread the thread, and I have some doubts: we have two options: (1) make the gait event format generic, and change it to an event.yaml
structure, or (2) keep `gaitEvent.yaml as it is, and add another format for other events.
You suggest going to option2, am I right?
I am proposing to merge them in only one format (yaml or csv), having specific labels for gait events.
Hum ok, so I would have to revise all repo and suggest using event.yaml filename.
File containing relevant labelled information