[三维重建] Unsupervised Learning of Probably Symmetric Deformable 3D Objects from Images in the Wild

转载请注明作者和出处： http://blog.csdn.net/john_bh/

论文链接： Unsupervised Learning of Probably Symmetric Deformable 3D Objects from Images in the Wild
论文链接：Arxiv 地址
作者及团队：牛津大学
会议及时间：CVPR 2020 best paper
code：作者开源github
主页：project 地址

1. 主要贡献

• Reconstructing human faces
  

• Reconstructing face paintings (model trained on CelebA real faces without fine-tuning)
  

• Reconstructing abstract faces (model trained on CelebA real faces without fine-tuning)
  

• Reconstructing cat faces
  

• Reconstructing synthetic cars
  

• Re-lighting effects
  

• Visualizing symmetry planes
  

• Visualizing asymmetry
  

• Results on video frames from VoxCeleb (applying model trained on CelebA faces frame-by-frame without fine-tuning)