SceneMover

This repository is the implementation of our SIGGRAPH Asia 2020 paper:

Scene Mover: Automatic Move Planning for Scene Arrangement by Deep Reinforcement Learning

Hanqing Wang, Wei Liang, Lap-Fai Yu.

Introduction

We propose a novel approach for automatically generating a move plan for scene arrangement. Given a scene like an apartment with many furniture objects, to transform its layout into another layout, one would need to determine a collision-free move plan.

Please refer to our paper for the detailed formulations.

Demonstration

Click here to watch the demonstration on Youtube.

Environment Installation

1. Install Requirements

   • python 3.6.9
   • g++ 5.4.0
   • CUDA 10.1
   • pillow 6.1.0
   • tensorflow 1.14.0
   • tensorboardx 1.8

2. Install Jupyter
   Install jupyter using the following scripts. pip install jupyter

3. Build Files

   cd src/utils
   g++ -shared -O2 search.cpp --std=c++11 -ldl -fPIC -o search.so

Q-Net Training

To be updated.

Inference:

To be updated.
<!— 1. Download the checkpoint of the agent to directory snap/agent/state_dict/best_val_unseen, the checkpoint of the speaker to directory snap/speaker/state_dict/best_val_unseen_bleu.

1. Start a Jupyter service and run the jupyter notebook evaluation.ipynb. —>

Contributors

To be updated.

TODO

• Release the checkpoint.
• Add training code.

Citation

Please cite this paper in your publications if it helps your research:

@article{wang2020scenem,
  author = {Hanqing Wang and Wei Liang and Lap-Fai Yu},
  title = {Scene Mover: Automatic Move Planning for Scene Arrangement by Deep Reinforcement Learning}, 
  journal = {ACM Transactions on Graphics},
  volume = {39},
  number = {6},
  year = {2020}
}

License

Scene Mover is freely available for non-commercial use, and may be redistributed under these conditions. Please see the license for further details. For commercial license, please contact the authors.

Contact Information

• hanqingwang[at]bit[dot]edu[dot]cn, Hanqing Wang