Support Website Contact Support Blog

Quality Report BUG - Wrong GCP Error values

I consistently find this error in the quality report or in the params files, in which the geolocation errors for the GCPs in the quality report are different from the errors calculated between estimated and measured coordinates.

For instance, I open the file in: ...\my_project\1_initial\params\my_project_measured_estimated_gcps_position.txt.

This file contains the measured coordinates for control points and also the estimated coordinates. The values are consistent with the measured coodinates I have input. The values are also consistent with the files my_project_estimated_gcps_position.txt and my_project_measured_gcps_position.txt.

Then I put these values in a spreadsheet and calculate the errors:
(I removed the initial digits of X and Y to protect the information)

Now, when I look at the quality report, I get this table (in which I inserted the checkpoints in the middle to make a single table).

image

Notice how all the “checkpoints” contain exactly the same error. However, the GCPs have big (and inconsistent) differences. Look at the highlighted cell in orange and compare to the others.

Below, the difference between the errors I calculated and the errors presented in the report:

image

What is going on here? Is Pix4D bugged for GPC errors?
What is wrong? The files? The quality report?

Pix4D Mapper Version: 4.5.6

Hi @Daniel_Moller,

Our apologies for the late response.

We will look into this issue and come back to you as soon as it is possible.

Regards

Hi @Daniel_Moller,

Those are two different statistics generated at different stages of the processing, and a small deviation between them is expected. Therefore, there is not a bug as the difference is smaller than the accuracy of the GCPs. Furthermore, I have to say that the param folder files are not supported and are for experimental use only.

Regards

Thank you for the answer, @Beata

I understand your explanation, but to be honest, that system behavior is very misleading for the user.
So the only numbers I can trust are the ones shown in the quality report?

Oi @Daniel_Moller,

I am sorry to hear that you found the system behaviour very misleading.

However, we make every effort to inform our clients about the situation, and for the params folder, it wasn’t any different. Our article What does the Output Params Folder contain? contains our disclaimer.

Indeed, the values shown in the Quality Report are correct, and those are the one to trust.

For your confidence, I asked our team to verify the statistics. They confirm correct functionality.

Let me know if there is anything else I can do for you.

Regards

Thank you, Beata.

I’ll just suggest an improvement then: to add the “user expected” file there (even though I understand this folder is experimental).
These files are very useful for making custom reports and further studies.

Hi @Daniel_Moller ,

Sorry, could you tell me what this user expected file should be? I think I don’t understand the idea.

Thanks!

For the user, two things are displayed in the software:

  • Measured position of the points (the positions input by the user)
  • Final calculated position of the points

These are the two files that would be great to have.

We have the “measured” file already.
But specifically for GCPs, the final calculated positions are not in the files. There is an “estimated” file that shows something that:

  • For checkpoints is the final position
  • For GCPs is not - we would expect that these were the final positions.

Hi @Daniel_Moller ,

Thank you for your feedback. I will pass it on to our team. :slight_smile:

Regards