Skip to content
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

Problem with Rigid Scenes #3

Open
arya-tama opened this issue Mar 22, 2021 · 6 comments
Open

Problem with Rigid Scenes #3

arya-tama opened this issue Mar 22, 2021 · 6 comments

Comments

@arya-tama
Copy link

arya-tama commented Mar 22, 2021

Hi,

I've been trying to use DefSLAM for dental model, which is a rigid object. My camera is already calibrated. However, once I start the camera using this command:

./DefSLAM ORBfile Yamlfile

It seems the Map Viewer threads try to find out how a single frame is deformed, instead of reconstructing scenes iteratively (see video). Can I use DefSLAM for completely rigid scenes? If so, what are steps to do? I try with Mandala datasets, and everything works fine. However, that example uses stereo images. I can't replicate this example as my camera is monocular.

https://drive.google.com/file/d/1Ou4pbnn_X5Ye5bREEz9hIh7e9vEqIWI8/view?usp=sharing

Best,
Arya

@JoseLamarca
Copy link
Collaborator

Hi Arya,

Thanks for trying our software! It seems a quite hard sequence. The algorithm works with rigid sequences too (always with enough parallax), the mandala only uses stereo for evaluation, there is a flag in the stereo settings that allows using the algorithm without evaluating it. I am going to check the executable only with monocular because it is being problematic.

In any case, I can see several potential issues coming from the algorithm itself for this sequence. The first is that the features of the scene seem really hard to track. The second is that the surface is discontinuous and not smooth what breaks the assumptions made. It starts as a plane and then converges to a similar shape. Unfortunately, I think it is far from the final application it was thought for. :S

The feature problem probably will be solved by the next version SD-DefSLAM. For the problem of discontinuities, we are working on it, but it is still a challenging problem.

I will let you know when I solve the issue with the monocular executable.
Best,

@JoseLamarca JoseLamarca pinned this issue Mar 23, 2021
@arya-tama
Copy link
Author

Hi Jose,

Thanks for your quick response!

I see, apparently this library works fine for certain sequences. Anyway, it would be great if you can show me how to work with monocular executable. I would be happy to share my testing result with different scenes later.

Best,
Arya

@JoseLamarca
Copy link
Collaborator

Sure! if it is possible, send me an example sequence that you want to process in particular to test it and I prepare the executable.

Best!

@JoseLamarca
Copy link
Collaborator

Hi Arya! did you get to correct the problem?

@ZJJ2022415
Copy link

Excuse me, have you solved the executable file problem of monocular camera yet?

@JoseLamarca
Copy link
Collaborator

Sorry, I could not reproduce the problem.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants