/FS2KToolbox

Primary LanguageMATLABMIT LicenseMIT

FS2KToolbox

FS2K Dataset

Towards the translation between Face <--> Sketch.

Download (photo+sketch+annotation): Google-drive, Baidu-disk, pw: FS2K.

For more details about the FS2K dataset, please visit this repo.

  • Put the source frames and result images as follows:
FS2K_PROJ_EVAL
├── photo
│       ├── photo1
│       ├── photo2
│       └── photo3
├── sketch
│       ├── sketch1
│       ├── sketch2
│       └── sketch3
├── I2S
│       ├── results_method1
│       ├── results_method2
│       └── ...
├── S2I
│       ├── results_method1
│       ├── results_method2
│       └── ...
├── anno_test.json
├── anno_train.json

Evaluation

  • Metric: SCOOT (designed for sketch evaluation) and SSIM.
  • Env: MATLAB (tested with MATLAB R2020b)
  • Run eval_FSS_results.m for testing.
  • Output: two excel files for saving SCOOT and SSIM scores of all methods: overall_scoot, scores_of_facial_parts.

vis_sample

Reference

@aticle{Fan2022FS2K,
  title={Deep Facial Synthesis: A New Challenge},
  author={Deng-Ping, Fan and Ziling, Huang and Peng, Zheng and Hong, Liu and Xuebin, Qin and Luc, Van Gool},
  journal={Machine Intelligence Research},
  year={2022}
}

@inproceedings{fan2019scoot,
  title={Scoot: A perceptual metric for facial sketches},
  author={Fan, Deng-Ping and Zhang, ShengChuan and Wu, Yu-Huan and Liu, Yun and Cheng, Ming-Ming and Ren, Bo and Rosin, Paul L and Ji, Rongrong},
  booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
  pages={5612--5622},
  year={2019}
}

@article{Wang2004ImageQA,
  title={Image quality assessment: from error visibility to structural similarity},
  author={Zhou Wang and Alan Conrad Bovik and Hamid R. Sheikh and Eero P. Simoncelli},
  journal={IEEE Transactions on Image Processing},
  year={2004},
  volume={13},
  pages={600-612}
}