Mysterious highway over North Sea?

Is this data error?

1 Like

You could have left out the question mark :joy:

There’s something quite wrong with his GEO tagging in some cases…

@laye Yes. This is an error with corrupt GPS. It has been filed and @peter is about to clean it up.

@jesolem, there are a lot of such cases on the map, just check the dots somewhere in the middle of the ocean, 50+% are there due to corrupt exif data / test uploads

p.s. I guess there should be some algos you can use to clean-up everything missing land surface beneath (with certain allowance, e.g. 1km to the nearest land surface)

There are always exceptions :joy:

This subtitle delivers irony in spades.

Most of these uploads are now gone, but since the camera is interleafed with other uploads by that user, it’s a tough task tracking these down. I’m doing it by deleting everything around 10km of York for the day of one of these uploads, have a bit left there.

2 Likes

Hey laye.
It is myself who has these incorrect locations currently being spotted around the United Kingdom.
I changed from an gopro hero 5 with GPS to a blackvue 590-2ch with GPS to capture my imagery.
For some reason the dashcam decides randomly to give some of the files an completely incorrect gpx track linked to the video footage (extremely incorrect Gps) I have no idea why.

I am currently trying to find an program I can run all the gpx tracks through to automatically detect incorrect files and just straight delete them off my pc and save Peter the troubles. But am grateful he has got a method to wipe the current ones uploaded incorrectly.

I am atm manually checking each gpx track before processing due to this issue but it is time consuming. So may save future uploads and capture projects until after investing in hopefully more reliable camera system.

Sorry my images have causes confusion and the troubles they may have causes mapillary to sort out my mess.

2 Likes

there are a lot of such cases on the map, just check the dots somewhere in the middle of the ocean, 50+% are there due to corrupt exif data / test uploads

1 Like