Mavic pro inaccurate layers

Hi,

I am trying to map forestry plantations in Malawi for an NGO (wellsforzoe.org) and so I am practising the mapping workflow in Germany.
In most of my tests the resulting map looks ok but is actually about 20-30m off…!

Even though the drona has good GPS signal… so 20-30m is quite a lot.
Do you have any tipps to improve this?

Thank you!
Kevin

Did you use ground control points (GCPs)? If not, the horizontal error could easily be 3-5m. But certainly not 20-30m. I suspect the topographic map is not very accurate either, or there is a projection issue. Can you overlay your ortho onto a different map, perhaps other imagery?

2 Likes

Could you please upload the quality report of your project?

▸ The quality report is located here (.pdf format): …\project_name\1_initial\report\project_name_report.pdf

Is the error of 20-30m in all directions? X,Y and Z?

The base layer map was open street maps, but with Google Earth satellite or Google maps layer it all is the same…
I did not use GCPs (and I also don’t know how the workflow would be for this either - but as I am testing mapping for a project in Malawi in the forest I believe it would be difficult to use GCPs.

Hello,

of course I can share this! maptest_report.pdf (1.6 MB)

I see that some images have not even been used - but do not really understand why. Is that indicated in the report somewhere? What could I do to increase quality?
(I was flying with 36km/h ISO 200, 1/240s, ~160 Above ground altitude) with Litchi

oh and I measured the distance of the error with a “measurement tool” in google earth pro - it has not been displayed by any software

Hello,

Thank you for sending the report.

From the quality report, it seems that the software respected the coordinates of your images. That being said, the software did not “move” your projects’ coordinates.
For more information, see the Geolocation Details section in the report

I suggest that you check the GPS coordinates of your images. Are the images projected well on the basemap of Pix4Dmapper?

For more information about how to improve the quality (not the 20-30m offset but the quality of the reconstruction) in dense vegetation areas:

I hope it helps :slight_smile:

adding manual tie points can also improve the reconstruction: