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 this image with the Dockerfile.
docker pull suhangpro/caffe:bpcn
- The docker image provided above uses CUDA 8, which is no longer supported if you have Volta GPUs (e.g. Titan V), Turing GPUs (e.g. RTX 2080), or newer ones. Adapting the Dockerfile to more recent GPUs should be straightforward—check out the example supporting up to Turing, courtesy of @zyzwhdx.
-
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
- 3D facade segmentation
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.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent