SSP3D: Semi-supervised single-view 3d reconstruction via prototype shape priors

@inproceedings{xing2022semi,
  title={Semi-supervised single-view 3d reconstruction via prototype shape priors},
  author={Xing, Zhen and Li, Hengduo and Wu, Zuxuan and Jiang, Yu-Gang},
  booktitle={ECCV},
  year={2022}
}

Set ShapeNet datasets

imagefile  dataset/ShapeNetRendering
voxelfile  dataset/ShapeNetVox32

Set Pix3D dataset

imagefile dataset/img/ voxelfile dataset/model/

train ShapeNet stage1

python runner_shapenet.py

train ShapeNet stage2

python runner_shapenet.py --finetune --weights=xxx.pth

test ShapeNet

python runner_shapenet.py --test --weights=xxx.pth

train Pix3D stage1

python runner_pix3d.py

train Pix3D stage2

python runner_pix3d.py --finetune --weights=xxx.pth

test Pix3D

python runner_pix3d.py --test --weights=xxx.pth