dpvo ---colmap

[li1u]

Hello, I am amazed by the effect of your article. But I have some questions to ask you.

First, when I use video data, the visualization result is good, but the stored result.ply file is not good.

Second, there are still some differences between the cameras.txt and other files in the result folder and the standard colmap data. I don’t know whether your code provides an interface for storing as standard colmap data.

[lahavlipson]

Unfortunately, I am not able to reproduce this issue. If you could point me to an example which reproduces the problem, that would be very helpful for me.

The --colmap flag saves a directory which can be opened using: colmap gui -> select file, import model -> select path/to/dpvo-output -> select no to only visualize reconstruction

When I do this on the demo from the colab notebook, I get:

FYI: The --colmap output does not contain information about the 2D feature tracks, since I assumed most people just wanted the camera poses and maybe a sparse reconstruction. If this would be helpful for people, I could add it.

[li1u]

Yes, it really works for me, hope you can add it.