@martin_ueding When I add a new segment, do I need to trigger something to get it matched against old activities?
3 I added got matched nicely, but one didn't....
#geoactivityplayground
@martin_ueding As far as I can tell I really should have plenty of activities that match the segment.

@LovesTha
I've seen something similar. I have the impression that this happens when the points are spaced to far apart. This happens especially with car rides. I would need to resample.

Could you see whether there is an open GitHub issue and add to that or open as new issue?

I'm not sure when I'll get to that, though. So better have it there such that it doesn't get lost.

Section matching not matching expected activities · Issue #413 · martin-ueding/geo-activity-playground

Basic metadata: V1.23.0 running in Docker importing GPX files from FittoTrack With this segment: and my heat map: I should have multiple activities matching the segment. Can give more data if needed.

GitHub
@martin_ueding yeah, the other sections have some rides only counting one direction and not the other, they are only up or down hill, and the down hill direction registers less often. The section with zero matches has a peak in the middle, so every run has some fast in it (only ~40kmph I'd guess)
@martin_ueding can it be points in the track being imported to be the segment has it's points too far apart?
@martin_ueding Is it terrible to have segments that are just a start and end gate?

@LovesTha
The current algorithm compares the points of the track with the points of the segment, not the line. That it something that will prevent a match if the points are too far apart. What I need to change is that the points of the track are compared to the interpolated line of the segment.

Problem currently is time and energy 😔

@martin_ueding so I should be able to get it to work by ensuring my segments have dense enough points...

(Maybe just reject importing segments that aren't dense enough)

@LovesTha
The solution is to interpolate automatically. Everything else is just a bad workaround.