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

Configuration for using ORB_SLAM with high resolution 12MP images. #168

Open
vik748 opened this issue Dec 13, 2018 · 0 comments
Open

Configuration for using ORB_SLAM with high resolution 12MP images. #168

vik748 opened this issue Dec 13, 2018 · 0 comments

Comments

@vik748
Copy link

vik748 commented Dec 13, 2018

Hi,
I have been trying to run ORB_SLAM on a set of high resolution images, but have been unable to get it configured to give any results. Here is some progress I have made:

  1. When I scale my 4000x3000 images down to 640x480, the algorithm works great without any issues. I also scale down the calibration images and use the resulting calibration. So this tells me that the image set and calibration is good. However, at full resolution, the algorithm isn't able to get enough track enough features to initialize the map.

  2. I have tried playing with all the combinations of the various parameters as follows:

  • nFeatures: I have tried increasing this up to 4,000, 10,000 ... 40,000. The system starts throwing errors if I make it 50,000 or more. The reasoning for the high number is that 12MP is approx 40 times area of 640x480. However no luck.
  • nLevels: Tried making this 10 but doesn't seem to help much.
  • fastTh: Have tied making it as low as 5 but still no map initialization.
  • nScoreType: Have tried both Harris and Fast but again no map initialization.

If anyone has any intuition as to what might help get the algorithm on track I am open to ideas. If someone really has the inclination to play with the dataset, I'd be happy to share.

Thanks in advance,
Vik

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

1 participant