Copyright (C) 2018 NVIDIA Corporation. All rights reserved. Licensed under the CC BY-NC-SA 4.0 license (https://creativecommons.org/licenses/by-nc-sa/4.0/legalcode).
@inproceedings{su18splatnet,
author={Su, Hang and Jampani, Varun and Sun, Deqing and Maji, Subhransu and Kalogerakis, Evangelos and Yang, Ming-Hsuan and Kautz, Jan},
title = {{SPLATN}et: Sparse Lattice Networks for Point Cloud Processing},
booktitle = {Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
pages = {2530--2539},
year = {2018}
}
-
Install Caffe and bilateralNN
Note that our code uses Python3.
- Please follow the instructions on the bilateralNN repo.
- A step-by-step installation guide for Ubuntu 16.04 is provided in INSTALL.md.
- Alternatively, you can install nvidia-docker and use this docker image:
You can also build your own image with this Dockerfile.
docker pull suhangpro/caffe:bpcn
-
Include the project to your python path so imports can be found, e.g.
export PYTHONPATH=<PATH_TO_PROJECT_ROOT>:$PYTHONPATH
-
Download and prepare data files under folder
data/
See instructions in data/README.md.
-
Usage examples
-
3D facade segmentation
- test pre-trained model
Prediction is output at
cd exp/facade3d ./dl_model_facade3d.sh # download pre-trained model SKIP_TRAIN=1 ./train_test.sh
pred_test.ply
, with evaluation results intest.log
. - or, train and evaluate
cd exp/facade3d ./train_test.sh
- test pre-trained model
-
ShapeNet Part segmentation
- test pre-trained model
Predictions are under
cd exp/shapenet3d ./dl_model_shapenet3d.sh # download pre-trained model ./test_only.sh
pred/
, with evaluation results intest.log
. - or, train and evaluate
cd exp/shapenet3d ./train_test.sh
- test pre-trained model
-
Joint 2D-3D experiments
(coming soon)
-
We make extensive use of bilateralNN, which is proposed in these publications:
- V. Jampani, M. Kiefel and P. V. Gehler. Learning Sparse High-Dimensional Filters: Image Filtering, Dense CRFs and Bilateral Neural Networks. CVPR, 2016.
- M.Kiefel, V. Jampani and P. V. Gehler. Permutohedral Lattice CNNs. ICLR Workshops, 2015.