-
Notifications
You must be signed in to change notification settings - Fork 1.7k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[softwares/SfM] Invalid -f input for SfM binaries -> Ceres local_parameterization check failed #727
Comments
Hi, I would like to understand if you are using the internal ceres version or an external one. |
Which file do you need ? I'm not sure where to get this information ... Is it the Makefile.cmake file ? |
It's the log output when you run cmake |
Here's the output :
|
Did you run the experiments by running the binary manually or by using the provided python scripts? |
I would suggest also to run the unit test to check what is going on on your computer |
I've a series of bash scripts that are calling directly the binaries. I use them also to manage a large data structure. I've recompiled the code with the test flag. Here's the output of
By the way (I'm not sure if it's relevant of useful), I've got many times the following warnings during compilation :
and
|
Test are ok, so I think the problem can perhaps comes from a special usage your are doing. I'm aware of the binder1st and binder2nd compilation warning and it is not related to your problem |
Seems like the error is due to a wrong configuration of the BA refinement settings inside the SfM_Global/Incremental binaries. Did you check the parameter -f you are using for the SfM tools? Now, we are using STRING based parameter, I think you are using the old configuration (that was using a boolean). Just removed the -f X option and I think everything will run fine |
@nils-hamel |
It was the problem ! I was still using the option of release 0.9. The two incremental and global pipelines works fine after passing the new -f parameter. So, problem solved ! Thanks for your help ! |
Thank you for your feedback. |
Hi @pmoulon,
I've compiled the v1.1 and made some verifications using the Sceaux Castle dataset. I got the following error (linked to ceres I assume) using both incremental and global pipelines :
local_parameterization.cc:103 Check failed: constant_parameters.size() < size Number of parameters held constant should be less than the size of the parameter block. If you wish to hold the entire parameter block constant, then a efficient way is to directly mark it as constant instead of using a LocalParameterization to do so.
Using the global pipeline, the error occurs after starting bundle adjustment. Using the incremental pipeline, the error also occurs during bundle adjustment as it refines the complete model (the resection ply file contains all the cameras which appears well placed).
Any idea what's going wrong there ?
OpenMVG 1.1 on Ubuntu 16.04
The text was updated successfully, but these errors were encountered: