chongshengzhang / Verisimilar-Image-Synthesis-for-Accurate-Detection-and-Recognition-of-Texts-in-Scenes

Code and data for the synthesis of scene text images.

Verisimilar-Image-Synthesis-for-Accurate-Detection-and-Recognition-of-Texts-in-Scenes

Sample datasets for scene text detection and recognition as described in the paper https://arxiv.org/abs/1807.03021

Detection:

1K: https://www.dropbox.com/s/5hirb8eal44ek4d/1K.tar.gz?dl=0

10K: https://www.dropbox.com/s/fu49xa4vqxhtrwm/10K.tar.gz?dl=0

Recognition:

1M: https://www.dropbox.com/s/7tgvrrp9p2u628r/1M.tar.gz?dl=0

The code and larger datasets will be released later.

Image Samples

Citation

If you find this project helpful for your research, please cite the following paper:

@article{fnzhan2018,
  author  = {Fangneng Zhan, Shijian Lu and Chuhui Xue},
  title   = {Verisimilar Image Synthesis for Accurate Detection and Recognition of Texts in Scenes},
  journal = {ECCV},
  pages   = {257-273},
  year    = {2018}
  }