Skip to content

Conversation

@hunminkim98
Copy link
Contributor

@hunminkim98 hunminkim98 commented Jan 28, 2026

Sorry for the confusion!
I've tested the AI agent and it produces the wrong PR.
I'll come back here when I've successfully implemented it!

- Add bundle_adjustment.py module with GPU-accelerated BA using ba-cuda
- Integrate BA into triangulation.py for post-triangulation refinement
- Add bundle_adjustment config section to Config.toml templates
- Add ba-cuda as optional dependency in pyproject.toml

Bundle Adjustment features:
- Levenberg-Marquardt optimization with analytical Jacobians
- Huber loss for outlier robustness
- Confidence-weighted optimization using keypoint scores
- Reference camera (cam 0) fixed during optimization
- Intrinsic parameters fixed, only extrinsics refined

Config options:
- triangulation.bundle_adjustment.enabled
- triangulation.bundle_adjustment.max_iterations
- triangulation.bundle_adjustment.tolerance
- triangulation.bundle_adjustment.huber_delta
- triangulation.bundle_adjustment.use_confidence_weights

Install with: pip install pose2sim[ba]
@hunminkim98 hunminkim98 deleted the feature/bundle-adjustment branch January 28, 2026 13:43
@davidpagnon
Copy link
Collaborator

Looking forward to it!

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

Successfully merging this pull request may close these issues.

2 participants