【CVPR 2019】DeepSDF: Learning Continuous Signed Distance Functions for Shape Representation
1. DeepSDF: Learning Continuous Signed Distance
Functions for Shape Representation
著者: J. J. Park, P. Florence, J. Straub,
R. Newcombe, S. Lovegrove
資料作成:千葉直也(東北大)
1
http://xpaperchallenge.org/cv
2. DeepSDF
• Signed Distance Function (SDF)を
直接学習することで三次元形状を
取り扱うネットワーク
– 形状をコード化して形状ごとのSDFを学習
– モデルサイズが小さい
– Codeの決め方に工夫(Auto-decoder)
16. 関連論文
16
• CVPR2019
– DeepSDF: Learning Continuous Signed Distance Functions for
Shape Representation
• Jeong Joon Park, Peter Florence, Julian Straub, Richard Newcombe,
Steven Lovegrove
– Occupancy Networks: Learning 3D Reconstruction in Function
Space
• Lars Mescheder, Michael Oechsle, Michael Niemeyer, Sebastian
Nowozin, Andreas Geiger
– Learning Implicit Fields for Generative Shape Modeling
• Zhiqin Chen, Hao Zhang
同時多発的に近いアイデアが出てきた
皆川さんの発表でも同じ論文をピックアップ
(いつも良質な資料をありがとうございます)
https://www.slideshare.net/takmin/20190706cvpr20193dshaperepresentation-153989245