This repository contains the source codes for the paper AtlasNet V2 - Learning Elementary Structures.

Théo Deprelle

Last update: Nov 11, 2022

Related tags

Overview

AtlasNet V2 - Learning Elementary Structures

This work was build upon Thibault Groueix's AtlasNet and 3D-CODED projects. (you might want to have a look at those)

This repository contains the source codes for the paper AtlasNet V2 - Learning Elementary Structures.

Citing this work

If you find this work useful in your research, please consider citing:

@inproceedings{deprelle2019learning,
  title={Learning elementary structures for 3D shape generation and matching},
  author={Deprelle, Theo and Groueix, Thibault and Fisher, Matthew and Kim, Vladimir and Russell, Bryan and Aubry, Mathieu},
  booktitle={Advances in Neural Information Processing Systems},
  pages={7433--7443},
  year={2019}
}

Project Page

The project page is available http://imagine.enpc.fr/~deprellt/atlasnet2/

Install

Clone the repo and install dependencies

This implementation uses Pytorch.

## Download the repository
git clone https://github.com/TheoDEPRELLE/AtlasNetV2.git
cd AtlasNetV2
## Create python env with relevant packages
conda create --name atlasnetV2 python=3.7
source activate atlasnetV2
pip install pandas visdom
conda install pytorch torchvision -c pytorch
conda install -c conda-forge matplotlib
# you're done ! Congrats :)

Training

Data

cd data; ./download_data.sh; cd ..

We used the ShapeNet dataset for 3D models.

When using the provided data make sure to respect the shapenet license.

The point clouds from ShapeNet, with normals go in data/customShapeNet
The corresponding normalized mesh (for the metro distance) goes in data/ShapeNetCorev2Normalized
the rendered views go in data/ShapeNetRendering

The trained models and some corresponding results are also available online :

The trained_models go in trained_models/

Build chamfer distance

The chamfer loss is based on a custom cuda code that need to be compile.

source activate pytorch-atlasnet
cd ./extension
python setup.py install

Start training

First launch a visdom server :

python -m visdom.server -p 8888

Check out all the options :

git pull; python training/train.py --help

Run the baseline :

git pull; python training/train.py --model AtlasNet --adjust mlp
git pull; python training/train.py --model AtlasNet --adjust linear

Run the Patch Deformation module with the different adjustment modules :

git pull; python training/train.py --model PatchDeformation --adjust mlp
git pull; python training/train.py --model PatchDeformation --adjust linear

Run the Point Translation module with the different adjustment modules:

git pull; python training/train.py --model PointTranslation --adjust mlp
git pull; python training/train.py --model PointTranslation --adjust linear

Monitor your training on http://localhost:8888/

Models

The models train on the SURREAL dataset for the FAUST competition can be found here

Acknowledgement

This work was partly supported by ANR project EnHerit ANR-17-CE23-0008, Labex Bezout, and gifts from Adobe to Ecole des Ponts.

License

MIT

Comments

Unable to download shapenet data

Hi,

I am trying to download data form download.sh script. But it is giving 404 error.

--2020-12-01 14:38:25-- https://cloud.enpc.fr/s/j2ECcKleA1IKNzk/download Resolving cloud.enpc.fr (cloud.enpc.fr)... 195.221.193.80 Connecting to cloud.enpc.fr (cloud.enpc.fr)|195.221.193.80|:443... connected. HTTP request sent, awaiting response... 404 Not Found 2020-12-01 14:38:26 ERROR 404: Not Found.

could you please provide an alternative link?

opened by brjathu 11
Question about evaluation critetion in paper?

Here, it is said that the reconstruction task is evaluated by chamfer distance. But for surreal data, the ground-truth correspondences are known. Why not just compute the L2 distance for correponding points?

opened by GostInShell 3
How to Generate 16384 points for Point Translation Module?

As discussed in https://github.com/ThibaultGROUEIX/AtlasNet/issues/42, I want to upsample the results of the point translation module. Since this module takes a fixed number of points into the network. I don't know whether training a new model taking 16384 points as input is justifiable to compare with our method.

opened by hzxie 2
The question about initialization of 'rand_grid' in the ./auxiliary/model.py

I have a question. In the file 'model.py', line 378,379, why the variable 'rand_grid' is initialized to uniform(0,1) before it is initialized to zero. What is the reason? Thanks!
bug

opened by tommaoer 2

Two bugs when running train.py

First bug is

Traceback (most recent call last):
  File "training/train.py", line 140, in <module>
    visdom = visdom.Visdom(env=opt.training_id, port=8888)
TypeError: __init__() got an unexpected keyword argument 'env'

and I delete env=opt.training_id, then i re-run this code. And Second bug is

Traceback (most recent call last):
  File "training/train.py", line 209, in <module>
    color =  [[125,125,125]]*(batch.size(1))
NameError: name 'batch' is not defined

opened by Yuzuki-N 0

unused model in PointTransLinAdj

It seems that a deformation layer is defined and not used. https://github.com/TheoDEPRELLE/AtlasNetV2/blob/master/auxiliary/model.py#L302

Did you intend to use this model?

opened by orenkatzir 0
About visualization

Hi, first thanks for your inspiring work! Point cloud rendering figures in your paper are beautiful as follows. How do you draw it? Using open3d, meshlab or other programmes?

Thanks!

opened by StevenZzz07 0
Pretrained Models

Hi,

I am trying to download data from https://cloud.enpc.fr/s/c27Df7fRNXW2uG3, but i get an 404 error. Could you please provide an alternative link? Thanks

opened by rspezialetti 0
The problem of test.

Dear professor, I have read the paper of " Learning Elementary Structures",and I have some problems. I have trained this network use datasets of Shapenet, and I get files of "network.pth" and "opt.pickle". But I can't find where is the "Elementary Structures" ,so I don't know how to compute correspondence use these "Elementary Structures". So I think your readme.md document is not complete, would you like to explain this issues.I don't know what to do after I finished trained my datasets, and how to get the correspondence. Looking for your early reply. Thank you!

opened by cainiaoshidai 0

Releases(1-beta)

1-beta(Sep 13, 2019)

Source code(tar.gz)
Source code(zip)

This repository contains the source codes for the paper AtlasNet V2 - Learning Elementary Structures.

Related tags

Overview

AtlasNet V2 - Learning Elementary Structures

Citing this work

Project Page

Install

Clone the repo and install dependencies

Training

Data

Build chamfer distance

Start training

Models

Acknowledgement

License

Comments

Releases(1-beta)

1-beta(Sep 13, 2019)

Owner

Théo Deprelle

This repository contains a set of codes to run (i.e., train, perform inference with, evaluate) a diarization method called EEND-vector-clustering.

This repository contains the source code and data for reproducing results of Deep Continuous Clustering paper

This repository contains the source code for the paper "DONeRF: Towards Real-Time Rendering of Compact Neural Radiance Fields using Depth Oracle Networks",

An image base contains 490 images for learning (400 cars and 90 boats), and another 21 images for testingAn image base contains 490 images for learning (400 cars and 90 boats), and another 21 images for testing

Implementation of the GVP-Transformer, which was used in the paper "Learning inverse folding from millions of predicted structures" for de novo protein design alongside Alphafold2

A weakly-supervised scene graph generation codebase. The implementation of our CVPR2021 paper ``Linguistic Structures as Weak Supervision for Visual Scene Graph Generation''

Code for the paper "Ordered Neurons: Integrating Tree Structures into Recurrent Neural Networks"

Source codes for the paper "Local Additivity Based Data Augmentation for Semi-supervised NER"

The source codes for ACL 2021 paper 'BoB: BERT Over BERT for Training Persona-based Dialogue Models from Limited Personalized Data'

Multiple paper open-source codes of the Microsoft Research Asia DKI group

This repository contains the code for the CVPR 2020 paper "Differentiable Volumetric Rendering: Learning Implicit 3D Representations without 3D Supervision"

This repository contains the source code of our work on designing efficient CNNs for computer vision

Using deep learning to predict gene structures of the coding genes in DNA sequences of Arabidopsis thaliana

This repository contains the code for the paper "Hierarchical Motion Understanding via Motion Programs"

This repository contains the code, data, and models of the paper titled "XL-Sum: Large-Scale Multilingual Abstractive Summarization for 44 Languages" published in Findings of the Association for Computational Linguistics: ACL 2021.

This repository contains a re-implementation of the code for the CVPR 2021 paper "Omnimatte: Associating Objects and Their Effects in Video."

This repository contains the code and models for the following paper.

This repository contains the official implementation code of the paper Improving Multimodal Fusion with Hierarchical Mutual Information Maximization for Multimodal Sentiment Analysis, accepted at EMNLP 2021.

This repository contains the PyTorch implementation of the paper STaCK: Sentence Ordering with Temporal Commonsense Knowledge appearing at EMNLP 2021.