Update README.md

This commit is contained in:
Xun Huang 2019-07-25 11:48:56 -07:00 committed by GitHub
parent 8bedce0be6
commit 470cfc4a77
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23

View file

@ -3,13 +3,16 @@
This repository contains a PyTorch implementation of the paper: This repository contains a PyTorch implementation of the paper:
[PointFlow : 3D Point Cloud Generation with Continuous Normalizing Flows](https://arxiv.org/abs/1906.12320). [PointFlow : 3D Point Cloud Generation with Continuous Normalizing Flows](https://arxiv.org/abs/1906.12320).
<br>
[Guandao Yang*](http://www.guandaoyang.com), [Guandao Yang*](http://www.guandaoyang.com),
[Xun Huang*](http://www.cs.cornell.edu/~xhuang/), [Xun Huang*](http://www.cs.cornell.edu/~xhuang/),
[Zekun Hao](http://www.cs.cornell.edu/~zekun/), [Zekun Hao](http://www.cs.cornell.edu/~zekun/),
[Ming-Yu Liu](http://mingyuliu.net/), [Ming-Yu Liu](http://mingyuliu.net/),
[Serge Belongie](http://blogs.cornell.edu/techfaculty/serge-belongie/), [Serge Belongie](http://blogs.cornell.edu/techfaculty/serge-belongie/),
[Bharath Hariharan](http://home.bharathh.info/) [Bharath Hariharan](http://home.bharathh.info/)
(* equal contribution)
<br>
ICCV 2019 (**Oral**)
## Introduction ## Introduction
@ -27,7 +30,7 @@ As 3D point clouds become the representation of choice for multiple vision and g
* G++ or GCC 5. * G++ or GCC 5.
* [PyTorch](http://pytorch.org/). Codes are tested with version 1.0.1 * [PyTorch](http://pytorch.org/). Codes are tested with version 1.0.1
* [torchdiffeq](https://github.com/rtqichen/torchdiffeq). * [torchdiffeq](https://github.com/rtqichen/torchdiffeq).
* (Optional) [Tensorboard](https://www.tensorflow.org/) for visualization of training process. * (Optional) [Tensorboard](https://www.tensorflow.org/) for visualization of the training process.
Following is the suggested way to install these dependencies: Following is the suggested way to install these dependencies:
```bash ```bash
@ -53,7 +56,7 @@ cd data
unzip ShapeNetCore.v2.PC15k.zip unzip ShapeNetCore.v2.PC15k.zip
``` ```
Please contact us if you need point clouds for ModelNet dataset. Please contact us if you need point clouds for the ModelNet dataset.
## Training ## Training
@ -71,7 +74,7 @@ Example training scripts can be found in `scripts/` folder.
## Pre-trained models and test ## Pre-trained models and test
Pretrained models can be downloaded from this [link](https://drive.google.com/file/d/1dcxjuuKiAXZxhiyWD_o_7Owx8Y3FbRHG/view?usp=sharing). Pretrained models can be downloaded from this [link](https://drive.google.com/file/d/1dcxjuuKiAXZxhiyWD_o_7Owx8Y3FbRHG/view?usp=sharing).
Following is the suggested way to evaluate the performance of the pre-trained models. The following is the suggested way to evaluate the performance of the pre-trained models.
```bash ```bash
unzip pretrained_models.zip; # This will create a folder named pretrained_models unzip pretrained_models.zip; # This will create a folder named pretrained_models
@ -87,7 +90,7 @@ CUDA_VISIBLE_DEVICES=0 ./scripts/shapenet_airplane_gen_test.sh
## Demo ## Demo
The demo relies on [Open3D](http://www.open3d.org/). Following is the suggested way to install it: The demo relies on [Open3D](http://www.open3d.org/). The following is the suggested way to install it:
```bash ```bash
conda install -c open3d-admin open3d conda install -c open3d-admin open3d
``` ```