HiFaceGAN: Face Renovation via Collaborative Suppression and Replenishment

Lingbo Yang1,3   Chang Liu2   Pan Wang3   Shanshe Wang1   Shanshe Wang1   Siwei Ma1   Wen Gao1

1Peking University  2University of Chinese Academy of Sciences  3Alibaba DAMO Academy 


Existing face restoration researches typically relies on either the degradation prior or explicit guidance labels for training, which often results in limited generalization ability over real-world images with heterogeneous degradations and rich background contents. In this paper, we investigate the more challenging and practical "dual-blind" version of the problem by lifting the requirements on both types of prior, termed as "Face Renovation"(FR). Specifically, we formulated FR as a semantic-guided generation problem and tackle it with a collaborative suppression and replenishment (CSR) approach. This leads to HiFaceGAN, a multi-stage framework containing several nested CSR units that progressively replenish facial details based on the hierarchical semantic guidance extracted from the front-end content-adaptive suppression modules. Extensive experiments on both synthetic and real face images have verified the superior performance of HiFaceGAN over a wide range of challenging restoration subtasks, demonstrating its versatility, robustness and generalization ability towards real-world face processing applications.

Quantitative Examples

Dynamic Perception-Distortion Tradeoff

Left The performances at 4x bicubic super resolution and blind restoration task on FFHQ dataset. ISPL dominates existing methods simultaneously on both aspects that it even reach beyond our initial hypothesis on the P-D boundary. Note the PSNR axis is inverted to better illustrate the tradeoff relationship.

Right: Super resolution results under varying scaling factors. Compared to explicit prior models, ISPL consistently delivers high-quality restoration results with rich facial details.


If you use our code or data, please cite:

                  title={HiFaceGAN: Face Renovation via Collaborative Suppression and Replenishment},
                  author={Lingbo Yang and C. Liu and P. Wang and Shanshe Wang and P. Ren and Siwei Ma and W. Gao},
                  journal={Proceedings of the 28th ACM International Conference on Multimedia},
Acknowledgements: page template comes from Po-Han Huang