Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox


Learning Flow-based Feature Warping for Face Frontalization with Illumination Inconsistent Supervision

Aug 16, 2020
Yuxiang Wei, Ming Liu, Haolin Wang, Ruifeng Zhu, Guosheng Hu, Wangmeng Zuo



Despite recent advances in deep learning-based face frontalization methods, photo-realistic and illumination preserving frontal face synthesis is still challenging due to large pose and illumination discrepancy during training. We propose a novel Flow-based Feature Warping Model (FFWM) which can learn to synthesize photo-realistic and illumination preserving frontal images with illumination inconsistent supervision. Specifically, an Illumination Preserving Module (IPM) is proposed to learn illumination preserving image synthesis from illumination inconsistent image pairs. IPM includes two pathways which collaborate to ensure the synthesized frontal images are illumination preserving and with fine details. Moreover, a Warp Attention Module (WAM) is introduced to reduce the pose discrepancy in the feature level, and hence to synthesize frontal images more effectively and preserve more details of profile images. The attention mechanism in WAM helps reduce the artifacts caused by the displacements between the profile and the frontal images. Quantitative and qualitative experimental results show that our FFWM can synthesize photo-realistic and illumination preserving frontal images and performs favorably against the state-of-the-art results.

* ECCV 2020. Code is available at: https://github.com/csyxwei/FFWM 


Share this with someone who'll enjoy it:

   Access Paper Source



Share this with someone who'll enjoy it: