Partial-View Object View Synthesis via Filtering Inversion

3DV 2024 (Highlight)

Official pytorch implementation

Fan-Yun Sun, Jonathan Tremblay, Valts Blukis, Kevin Lin, Danfei Xu, Boris Ivanovic, Peter Karkus, Stan Birchfield, Dieter Fox, Ruohan Zhang, Yunzhu Li, Jiajun Wu, Marco Pavone, Nick Haber

Usage

1. environmental setup

$ pip install -r requirements.txt

2. dataset preparation (see below)

3. refer to the sample command below:

# Scannet
python main.py scene_id=scene0038_00

Data Preparation

Run prepare_data.sh. The resulting directory should have the following file structure:

|-- data/
|   |-- scannet/
|   |   |-- scans/
|   |   |-- scan2cad/
|   |   |-- processed_scannet/
|   |-- gt_object_mesh/

TODO

[ ] parallelize the inversion process

Broader Information

FINV builds upon several previous works:

• PTI: Pivotal Tuning for Latent-based editing of Real Images
• Efficient Geometry-aware 3D Generative Adversarial Networks
• GET3D: A Generative Model of High Quality 3D Textured Shapes Learned from Images

Citation

@article{sun2023partial,
  title={Partial-View Object View Synthesis via Filtered Inversion},
  author={Sun, Fan-Yun and Tremblay, Jonathan and Blukis, Valts and Lin, Kevin and Xu, Danfei and Ivanovic, Boris and Karkus, Peter and Birchfield, Stan and Fox, Dieter and Zhang, Ruohan and others},
  journal={International Conference on 3D Vision (3DV)},
  year={2024}
}