Accuracy in the quality report

My request concerns the accuracy of the model. I did a lot of research but I didn’t find anything that reassures me 100%.

I understood that precision can be absolute and relative. Now I wanted to know if these values are present directly in the quality report or if they should be calculated with the GSD. If they are present in the quality report where?
For me the most telling values are the “mean reprojection error” and the RMS error of the GCP. But are these values really reliable indicators?

I attach a quality report of a project so that you give me directly the values to take into account so that I understand better.
Dardennes - Mavic mini - nadir - GCP_report.pdf (201.2 KB)

Thank you in advance for your help,


Hello @marg.qui
Thank you for sharing the quality report with us.
To answer your question about where to find the absolute and relative geolocation variance, you can find them at these parts in the quality report.

After taking a look at the quality report I noticed a couple of issues regarding the 1) Initial Image Position and the yellow triangle warning sign beside 2) Camera Optimization under Quality Check on the first page.

Fundamentally speaking, we need enough images and enough overlap to produce a good representation depending on each project. We recommend at least 75% frontal overlap (with respect to the flight direction) and at least 60% side overlap (between flying tracks). It is recommended to take the images with a regular grid pattern (Figure 1).

Step 1. Before Starting a Project > 1. Designing the Image Acquisition Plan > a. Selecting the Image Acquisition Plan Type

I would suggest taking a look at this article and increasing the overlap of images by working towards the recommended flight line for your area of interest.

Following the guidelines presented above will get you started in the right direction.

Before we jump into processing your new dataset with the appropriate number of images and overlap, I also suggest to get a few more GCPs. You can set a couple of GCPs as checkpoints as well. are used to assess the absolute accuracy of the model. The marks of the checkpoints are used to estimate its 3D position as well as potential errors in the clicks. This way, the relative accuracy of the area of the checkpoints may improve.

If the Camera Optimization under Quality Check is still greater than 5% with the processing option you previously selected, I recommend taking a look at this article Quality Report Help, and follow the below,

  • Process with lower Keypoints Image Scale: i.e. instead of full, choose half
  • Enable Geometrically Verified Matching
  • Set “internal calibration parameters” to All Prior in Step 1 processing options.

I hope this information helps. Feel free to share with us how it goes.

Best regards,

P1_manesar_report.pdf (2.2 MB)
Dear Rosana_Lin help me to figure out, is my data has accuracy or not, also i want to tell you that, we did not use the check points. so kindly help us to find the accuracy of the data and what is the meaning of red ink points in the report under the section of “Absolute Geolocation Variance”

Firstly, I have images with EGM 96 Geoid ( auto detected by images) and GCP’s with EGM2008 Geoid, Does it affect in the accuracy while using that GCP’s values.
Secondly, Does the elevation changes according to the EGM2008 Geoid or Creates a problem in accuracy while using two EMG Geoids