MingyuY/deformable-view-synthesis

Official Implementation for "Novel View Synthesis on Unpaired Data by Conditional Deformable Variational Auto-Encoder"(ECCV2020)

deformable-view-synthesis (ECCV2020)

Tensorflow implementation of Novel View Synthesis on Unpaired Data by Conditional Deformable Variational Auto-Encoder.

The network architecture

Experiment Results

• chair

• interpolation

Preparation

• Prerequisites
  • Tensorflow
  • Python 2.x with matplotlib, numpy and scipy
• Dataset
  • 3D chair
  • MultiPIE
  • Images should be placed in ./datasets/
• Download model
  • Tool model
    • model should be placed in ./models/

Quick Start

Exemplar commands are listed here for a quick start.

dataset

• prepare dataset

  python dataset.py --path 'Path to Training Dataset'
  

Training

• To train with size of 128 X 128

  python cdvae.py --mode training

Testing

• Example of generating samples

  python cdvae.py --mode generation --batch_size 1 --model_path 'Path to Training Model'

Citation

If this work is useful for your research, please consider citing:

@inproceedings{yin2020novel,
 title={Novel View Synthesis on Unpaired Data by Conditional Deformable Variational Auto-Encoder},
 author={Yin, Mingyu and Sun, Li and Li, Qingli},
 booktitle={European Conference on Computer Vision},
 pages={87--103},
 year={2020},
 organization={Springer}
 }