No description, website, or topics provided.
Clone or download
Latest commit d02dc89 Sep 20, 2018
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
pointnet2 first commit Aug 23, 2018
.gitignore Initial commit Aug 23, 2018
LICENSE Initial commit Aug 23, 2018
README.md Update README.md Sep 21, 2018
data_prep.py add evaluation code Sep 3, 2018
evaluation.py add evaluation code Sep 3, 2018
model.py add evaluation code Sep 3, 2018
pnmem_cell.py first commit Aug 23, 2018
seg_util.py add evaluation code Sep 3, 2018
train.py add evaluation code Sep 3, 2018

README.md

Deep Part Induction from Articulated Object Pairs

Introduction

This work is based on our SIGGRAPH Asia 2018 paper. You can find the arXiv version of the paper here. In this repository, we release the training and evaluation code, data as well as the pre-trained models.

To Get Start

Download training and validation data from

https://shapenet.cs.stanford.edu/ericyi/data_partmob.zip

Compile the PointNet++ code in "pointnet2"

Train the corrspondence proposal and the flow module through

python train.py --stage 1

Train the hypothesis generation and the verification submodule through

python train.py --stage 2

Train the hypothesis selection submodule through

python train.py --stage 3

Evaluate the model through

python evaluation.py

You can also download the pretrained model from the following link:

https://shapenet.cs.stanford.edu/ericyi/pretrained_model_partmob.zip

License

Our code and data are released under MIT License (see LICENSE file for details).