/DIB-R

Learning to Predict 3D Objects with an Interpolation-based Differentiable Renderer (NeurIPS 2019)

Primary LanguagePythonMIT LicenseMIT

DIB-Render

This is the official inference code for:

Learning to Predict 3D Objects with an Interpolation-based Differentiable Renderer (NeurIPS 2019)

Wenzheng Chen, Jun Gao*, Huan Ling*, Edward J. Smith*, Jaakko Lehtinen, Alec Jacobson, Sanja Fidler

[Paper] [Project Page]

Usage

Install dependencies

This code requires PyTorch 1.1.0 and python 3+. Please install dependencies by

pip install -r requirments.txt

Compile the DIB-Render

cd dib-render/cuda_dib_render
python build.py install

Inference

python test-all.py \
 --g_model_dir ./checkpoints/g_model.pth \
 --svfolder ./prediction \
 --data_folder ./dataset \
 --filelist ./test_list.txt

To get the evaluation IOU, please first download the tool Binvox and install it's dependencies,

Voxelize the prediction using Binvox

python voxelization.py  --folder ./prediction

To evaluate the IOU, please first install binvox-rw-py following this Link, then run the script

python check_iou.py --folder ./prediction  --gt_folder ./dataset 

To get the boundary F-score, please run the following script

python check_chamfer.py --folder ./prediction  --gt_folder ./dataset 

Ciatation

If you use the code, please cite our paper:

@inproceedings{chen2019dibrender,
title={Learning to Predict 3D Objects with an Interpolation-based Differentiable Renderer},
author={Wenzheng Chen and Jun Gao and Huan Ling and Edward Smith and Jaakko Lehtinen and Alec Jacobson and Sanja Fidler},
booktitle={Advances In Neural Information Processing Systems},
year={2019}
}