Skip to content
Project code for "Direct Fitting of Gaussian Mixture Models"
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
bunny
mixture
.gitignore
LICENSE
README.md
cube-covariance test.ipynb
format_graphs-Copy1.ipynb
format_graphs.ipynb
gen_gmm.ipynb
gmm_fit.py
gmm_fit2.py
gmm_fit3.py
gmm_fit_extra.py
graph_scales.ipynb
ground_truth_testing.ipynb
icp_test.ipynb
likelihood.ipynb
llcomp.py
plot_trajec.ipynb
reg_results.ipynb
reg_viz-extra.ipynb
reg_viz.ipynb
registration_test.py
registration_test_extra.py
render_tmp.py
road_graphic.py
tri_test.py
tri_verts_graph-Copy1.ipynb
tri_verts_graph.ipynb
vis_fig.py
vis_fitting.py
vis_fitting_bunny.py
vis_fitting_bunny_mesh.py
visualize gmm.ipynb

README.md

direct_gmm

This is the source code and project history for the following publication

Direct Fitting of Gaussian Mixture Models by Leonid Keselman and Martial Hebert (arXiv version here)

Overview

Almost all files used in the development and testing of this project are in this folder. The data files for the Stanford Bunny is included in bunny.

  • mixture contains the modifed version of scikit-learn with the proposed techniques.
  • gmm_fit.py and gmm_fit2.py contain the two sets of the bunny likelihood experiments
  • registration_test.py contains the mesh registration (P2D) experiments
  • Files with _extra are usually just copies for non-Stanford Bunny experiments
  • gen_gmm.ipynb and gen_gmm_mine.ipynb generate GMM models from the TUM dataset, with and without uncertainty models
  • reg_results.ipynb performs D2D registration between the GMM models built from the TUM dataset.
You can’t perform that action at this time.