Skip to content

A minimal PyTorch implementation of Bundle Adjustment

Notifications You must be signed in to change notification settings

Cufix/micro-bundle-adjustment

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

18 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Micro Bundle Adjustment 🌱

Getting into bundle adjustment solvers can be hard. To make it easier, I have made a very basic pure Pytorch implementation, useful for teaching/learning. Despite being less than 100 lines of code, it can handle 10^6 3D points with ease, due to utilizing sparsity and GPU.

Features

  • Basic Implementation of two-view bundle adjustment for any type of camera

TODO List

  • Example with residual function
  • Generalize implementation to K-views
  • [1/2] Clean up and document code
  • []

Usage

See demos.

Basically you define a (non-batched) function for your residuals, which takes in your observations, camera params, and 3D points and outputs the residual. Sending this together with an initial guess is all you need to do! Gradients are computed automatically, and you can use arbitrary camera models.

About

A minimal PyTorch implementation of Bundle Adjustment

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%