Skip to content

Latest commit

 

History

History
72 lines (53 loc) · 2.15 KB

README.md

File metadata and controls

72 lines (53 loc) · 2.15 KB

Text2img-tf2

A Tensorflow-2.0 version implementation of Stackgan++.
Han Zhang, Tao Xu, Hongsheng Li, Shaoting Zhang, Xiaogang Wang, Xiaolei Huang, Dimitris N. Metaxas: StackGAN++: Realistic Image Synthesis with Stacked Generative Adversarial Networks in TPAMI 2019. arxiv

Architecture

Dependencies

Datasets

We use images in Oxford 102 flower as our training and testing data. For captions, you can download with this URL. To turn the text captions into embedding, we utilize skip-thoughts model. You can find more details in the repo.

Installation

git clone https://github.com/jaroslaw1007/Text2img-tf2.git

Training

First you need to go config.py to fill in the directories you place the images, text captions, dictionaries, lookup table.

cfg.DATASET.IMAGE_PATH = ...
cfg.DATASET.CAPTION_PATH = ...
cfg.DATASET.LOOK_UP = ...
cfg.DATASET.DICTIONARY_PATH = ...

Then

python main.py

Demo

Citing

@article{DBLP:journals/pami/ZhangXLZWHM19,
  author    = {Han Zhang and
               Tao Xu and
               Hongsheng Li and
               Shaoting Zhang and
               Xiaogang Wang and
               Xiaolei Huang and
               Dimitris N. Metaxas},
  title     = {StackGAN++: Realistic Image Synthesis with Stacked Generative Adversarial
               Networks},
  journal   = {{IEEE} Trans. Pattern Anal. Mach. Intell.},
  volume    = {41},
  number    = {8},
  pages     = {1947--1962},
  year      = {2019},
  url       = {https://doi.org/10.1109/TPAMI.2018.2856256},
  doi       = {10.1109/TPAMI.2018.2856256},
  timestamp = {Fri, 26 Feb 2021 08:54:53 +0100},
  biburl    = {https://dblp.org/rec/journals/pami/ZhangXLZWHM19.bib},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}