Compositionally Generalizable 3D Structure Prediction

In this work, We bring in the concept of compositional generalizability and factorizes the 3D shape reconstruction problem into proper sub-problems, each of which is tackled by a carefully designed neural sub-module with generalizability guarantee. Experiments on PartNet show that we achieve superior performance than baseline methods, which validates our problem factorization and network designs. Link to our paper.

Check our YouTube videos below for more details.

If you find this project useful for your research, please cite:

@article{han2020compositionally,
author = {Han, Songfang and Gu, Jiayuan and Mo, Kaichun and Yi, Li and Hu, Siyu and Chen, Xuejin and Su, Hao},
title = {{C}ompositionally {G}eneralizable 3{D} {S}tructure {P}rediction},
journal = {arXiv preprint},
year = {2020}}

How to use

Installation

Check out the source code

git clone https://github.com/hansongfang/CompNet.git && cd CompNet
Install dependencies

conda env create -f environment.yml && conda activate CompNet
Compile CUDA extensions

cd common_3d && bash compile.sh