Use GPS time for videos if the difference with creation time is too big #659
Conversation
malconsei
force-pushed
the
feat-gps-time-2
branch
from
e7ff738 to
4ca41e9
Compare
| continue | ||
| track.append( | ||
| geo.PointWithFix( | ||
| geo.GpsPoint( |
There was a problem hiding this comment.
Here I guess we can keep using geo.Point as unix_timestamp == time?
| epoch_ms = int(match.group(1)) | ||
| yield geo.Point( | ||
| yield geo.GpsPoint( | ||
| time=epoch_ms, |
There was a problem hiding this comment.
It seems to be a bug here: Point.time should be always seconds (either seconds since epoch or offsets in seconds).
Could you change it to time=epoch_ms / 1000 and remove the / 1000 at L107?
| lon=nmea.longitude, | ||
| alt=nmea.altitude, | ||
| angle=None, | ||
| unix_timestamp=int(epoch_ms / 1000), |
There was a problem hiding this comment.
Doesn't need as unix_timestamp == time
| alt=box.data.altitude, | ||
| angle=None, | ||
| unix_timestamp=int(_gps_timestamp_to_unix(box.data.time_gps_epoch)), | ||
| gps_fix=geo.GPSFix(box.data.gps_fix_type), |
There was a problem hiding this comment.
Adding gps_fix/gps_ground_speed/gps_horizontal_accuracy here will trigger gps filtering which affect GPS points and video slicing. Let's do it in a separate PR with proper testing to be safe.
| for some info on GPS timestamp. Here we arbitrarily use the offset as of 2023, since we | ||
| are not interested in precision to the second. | ||
| """ | ||
| return ts + 315964800 - 18 |
| def _gps_timestamp_to_unix(ts: float) -> float: | ||
| """ | ||
| Convert a GPS timestamp to UNIX timestamp. | ||
| See https://stackoverflow.com/questions/20521750/ticks-between-unix-epoch-and-gps-epoch |
| def filter_points_by_elst( | ||
| points: T.Iterable[geo.Point], elst: T.Sequence[T.Tuple[float, float]] | ||
| ) -> T.Generator[geo.Point, None, None]: | ||
| points: T.Iterable[geo.GpsPoint], elst: T.Sequence[T.Tuple[float, float]] |
There was a problem hiding this comment.
Seems no need to be GpsPoint here: can cast GpsPoint to Point to use this function.
| alt=point.elevation, | ||
| angle=None, | ||
| gps_ground_speed=point.speed, | ||
| unix_timestamp=int(time), |
There was a problem hiding this comment.
no need unix_timestamp here as time == unix_timestamp
| lon=lon, | ||
| alt=alt, | ||
| angle=None, | ||
| unix_timestamp=int(time), |
| gps_precision: T.Optional[float] = None | ||
| gps_ground_speed: T.Optional[float] = None | ||
| gps_horizontal_accuracy: T.Optional[float] = None | ||
| unix_timestamp: T.Optional[int] = None |
There was a problem hiding this comment.
Using float instead of int looks more consistent as time uses float (although it ended being rounded as video created time)
malconsei
force-pushed
the
feat-gps-time-2
branch
from
4ca41e9 to
57e853d
Compare
|
LGTM. The only concern is the ground speed added in |
|
Hi @malconsei! Thank you for your pull request. We require contributors to sign our Contributor License Agreement, and yours needs attention. You currently have a record in our system, but the CLA is no longer valid, and will need to be resubmitted. ProcessIn order for us to review and merge your suggested changes, please sign at https://code.facebook.com/cla. If you are contributing on behalf of someone else (eg your employer), the individual CLA may not be sufficient and your employer may need to sign the corporate CLA. Once the CLA is signed, our tooling will perform checks and validations. Afterwards, the pull request will be tagged with If you have received this in error or have any questions, please contact us at cla@meta.com. Thanks! |
|
Keeping an eye on this one. It would great to have GPS time set as the default to avoid capture time data issues when the camera resets its clock. This is a common issue on GoPros. |
Some cameras reset their internal clock when the battery runs out, or otherwise get out of sync with the world clock.