/BEGAN-CS

Escaping from Collapsing Modes in a Constrained Space (ECCV 2018)

Primary LanguagePython

Escaping from Collapsing Modes in a Constrained Space

Chia-Che Chang*, Chieh Hubert Lin*, Che-Rung Lee, Da-Cheng Juan, Wei Wei, Hwann-Tzong Chen

The authors' TensorFlow implementation of ECCV'18 paper, "Escaping from Collapsing Modes in a Constrained Space".

** This is not an official Google product **

How to Use

    # download CelebA dataset
    python download.py
    
    # Training: 
    ./train.sh
   
    # Testing: 
    ./test.sh

PCA Visualization

Model Architecture

Results

Train on CelebA dataset


Train on 1/10 CelebA dataset


Selected disentangled representations of BEGAN-CS


Two-dimensional combinations of disentangled representations


Experimental results on the synthetic dataset


Image reconstruction results


Interpolation

Citation

@InProceedings{Chang_2018_ECCV,
author = {Chang, Chia-Che and Hubert Lin, Chieh and Lee, Che-Rung and Juan, Da-Cheng and Wei, Wei and Chen, Hwann-Tzong},
title = {Escaping from Collapsing Modes in a Constrained Space},
booktitle = {The European Conference on Computer Vision (ECCV)},
month = {September},
year = {2018}
}

Reference

  1. syentic dataset
  2. BEGAN
  3. tensorflow-generative-model-collections