Replies: 2 comments 1 reply
-
Thanks for continuing the discussion. Yes, I believe that the CSV file solution is good! How would text fields be processed? (e.g., overlaying a ML classifier output on the video) Another thing, how would you handle unsynced data. For instance one sensor output data at 20 Hz, while the other one at 1Hz. |
Beta Was this translation helpful? Give feedback.
-
Hadn't really thought of any of these things, hence the discussion. :-) I do want to keep things as simple as possible.. right now we don't support text at all within the metrics system, so that might not make the first iteration of such a capability... There are two approaches that I can think of for multiple sensors...1) repeat the sensor data as required across time 2) allow multiple input files. For 1 the creator of the CSV file would just repeat the value as required for each datetime, so a 1Hz value would be just repeated 20 times alongside a 20Hz value. This might be more difficult for people to create though.
or just two files.
|
Beta Was this translation helpful? Give feedback.
-
Following on from #71 - It might be useful to be able to import timeseries data into gopro-overlay so that it can be added to widgets, same as information from a GPX file or GoPro video.
Having had a 2 second think about this - could be terrible idea - it occurred to me that a simple csv file could be used.
We already support access to metrics by name, so accessing arbitrary fields would be relatively straightforward - you'd just have to reference the field in the "metric"
All numeric fields already support time-based interpolation, and these would continue to be interpolated, just as they are now. Interpolation might not be suitable for some fields, but if required we could address that in a different change.
The header row from the CSV file would be used to give fields names, with a required datetime field. Units could be added as per the example.
datetime, field1 (unit), field2 (unit)
2011-12-01T19:03:17.902,1234.5,2325.0
...
I think this might be 1) easy to implement 2) compatible with wide number of sensors
Any thoughts?
Beta Was this translation helpful? Give feedback.
All reactions