point_cloud_completion

Point Cloud Completion of Foot Shape From a Single Depth Map for Fit Matching Using Deep Learning View Synthesis

Nolan Lunscher and John Zelek

This code was used for my paper presented in the Computer Vision for Fashion Workshop in ICCV 2017. If you find it useful in your works, please site the original paper.

This project and parts of its code were inspired by "Multi-view 3D Models from Single Images with a Convolutional Network", which you may find useful as well: https://github.com/lmb-freiburg/mv3d

Dependencies

This program has the below dependencies but will likely work with newer versions as well:

• Ubuntu 16.04
• Python 2.7
• Tensorflow 0.12
• Panda3D 1.9.3
• MatLab 2016b
• Opencv 3.4

Usage

1. Download: caesar-norm-wsx-fitted-meshes (http://humanshape.mpi-inf.mpg.de/)
   • place in "../Data"
2. Run: mat_model_mesh_centerfoot.m from inside dataset_utils
   • make sure mat_folder is set correctly
3. Run: convert_objs2bam.sh from inside dataset_utils
   • make sure obj_folder and bam_folder are set correctly
4. Run: pc_completion_make_data.py
   • Run for train, test and val
   • make sure to set the appropriate values in the file for each set
5. Run: pc_completion.py - train
   • use run_option = 1
   • make sure self.dataset_folder is set correctly
6. Run: pc_completion.py - test depth map
   • use run_option = 3
   • make sure self.dataset_folder is set correctly
   • this will provide the depth map error
7. Run: pc_errors.m from inside dataset_utils - test point cloud
   • make sure that folder, obj_nums and im_nums as set correctly

License and Citation

All code is provided for research purposes only and without any warranty. Any commercial use requires our consent. When using the code in your research work, please cite the following paper:

@InProceedings{Lunscher_2017_ICCV_Workshops,
author = {Lunscher, Nolan and Zelek, John},
title = {Point Cloud Completion of Foot Shape From a Single Depth Map for Fit Matching Using Deep Learning View Synthesis},
booktitle = {The IEEE International Conference on Computer Vision (ICCV) Workshops},
month = {Oct},
year = {2017}
}