search

chrischoy / 3D-R2N2

Single/multi view image(s) to voxel reconstruction using a recurrent neural network
MIT License
1.36k stars 295 forks source link

readme

3D-R2N2: 3D Recurrent Reconstruction Neural Network

This repository contains the source codes for the paper Choy et al., 3D-R2N2: A Unified Approach for Single and Multi-view 3D Object Reconstruction, ECCV 2016. Given one or multiple views of an object, the network generates voxelized ( a voxel is the 3D equivalent of a pixel) reconstruction of the object in 3D.

Citing this work

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

@inproceedings{choy20163d,
  title={3D-R2N2: A Unified Approach for Single and Multi-view 3D Object Reconstruction},
  author={Choy, Christopher B and Xu, Danfei and Gwak, JunYoung and Chen, Kevin and Savarese, Silvio},
  booktitle = {Proceedings of the European Conference on Computer Vision ({ECCV})},
  year={2016}
}

News

Project Page

The project page is available at http://cvgl.stanford.edu/3d-r2n2/.

Overview

Overview Left: images found on Ebay, Amazon, Right: overview of 3D-R2N2

Traditionally, single view reconstruction and multi-view reconstruction are disjoint problems that have been dealt using different approaches. In this work, we first propose a unified framework for both single and multi-view reconstruction using a 3D Recurrent Reconstruction Neural Network (3D-R2N2).

3D-Convolutional LSTM 3D-Convolutional GRU Inputs (red cells + feature) for each cell (purple)
3D-LSTM 3D-GRU 3D-LSTM

We can feed in images in random order since the network is trained to be invariant to the order. The critical component that enables the network to be invariant to the order is the 3D-Convolutional LSTM which we first proposed in this work. The 3D-Convolutional LSTM selectively updates parts that are visible and keeps the parts that are self-occluded.

Networks We used two different types of networks for the experiments: a shallow network (top) and a deep residual network (bottom).

Results

Please visit the result visualization page to view 3D reconstruction results interactively.

Datasets

We used ShapeNet models to generate rendered images and voxelized models which are available below (you can follow the installation instruction below to extract it to the default directory).

Installation

The package requires python3. You can follow the direction below to install virtual environment within the repository or install anaconda for python 3.

git clone https://github.com/chrischoy/3D-R2N2.git
cd 3D-R2N2
conda create -n py3-theano python=3.6
source activate py3-theano
conda install pygpu
pip install -r requirements.txt
cp .theanorc ~/.theanorc

Running demo.py

sudo apt-get install meshlab
python demo.py prediction.obj

The demo code takes 3 images of the same chair and generates the following reconstruction.

Image 1 Image 2 Image 3 Reconstruction 
deactivate

Training the network

source py3/bin/activate
mkdir ShapeNet/
wget http://cvgl.stanford.edu/data2/ShapeNetRendering.tgz
wget http://cvgl.stanford.edu/data2/ShapeNetVox32.tgz
tar -xzf ShapeNetRendering.tgz -C ShapeNet/
tar -xzf ShapeNetVox32.tgz -C ShapeNet/
./experiments/script/res_gru_net.sh

Note: The initial compilation might take awhile if you run the theano for the first time due to various compilations. The problem will not persist for the subsequent runs.

Using cuDNN

To use cuDNN library, you have to download cuDNN from the nvidia website. Then, extract the files to any directory and append the directory to the environment variables like the following. Please replace the /path/to/cuDNN/ to the directory that you extracted cuDNN.

export LD_LIBRARY_PATH=/path/to/cuDNN/lib64:$LD_LIBRARY_PATH
export CPATH=/path/to/cuDNN/include:$CPATH
export LIBRARY_PATH=/path/to/cuDNN/lib64:$LD_LIBRARY_PATH

For more details, please refer to http://deeplearning.net/software/theano/library/sandbox/cuda/dnn.html

Follow-up Paper

Gwak et al., Weakly supervised 3D Reconstruction with Adversarial Constraint, project website

Supervised 3D reconstruction has witnessed a significant progress through the use of deep neural networks. However, this increase in performance requires large scale annotations of 2D/3D data. In this paper, we explore inexpensive 2D supervision as an alternative for expensive 3D CAD annotation. Specifically, we use foreground masks as weak supervision through a raytrace pooling layer that enables perspective projection and backpropagation. Additionally, since the 3D reconstruction from masks is an ill posed problem, we propose to constrain the 3D reconstruction to the manifold of unlabeled realistic 3D shapes that match mask observations. We demonstrate that learning a log-barrier solution to this constrained optimization problem resembles the GAN objective, enabling the use of existing tools for training GANs. We evaluate and analyze the manifold constrained reconstruction on various datasets for single and multi-view reconstruction of both synthetic and real images.

License

MIT License