HyperCube: Implicit Field Representations of Voxelized 3D Models

Authors: Magdalena Proszewska, Marcin Mazur, Tomasz Trzcinski, Przemysław Spurek

[Paper]

Abstract

Recently introduced implicit field representations offer an effective way of generating 3D object shapes. They leverage implicit decoder trained to take a 3D point coordinate concatenated with a shape encoding and to output a value which indicates whether the point is outside the shape or not. Although this approach enables efficient rendering of visually plausible objects, it has two significant limitations. First, it is based on a single neural network dedicated for all objects from a training set which results in a cumbersome training procedure and its application in real life. More importantly, the implicit decoder takes only points sampled within voxels (and not the entire voxels) which yields problems at the classification boundaries and results in empty spaces within the rendered mesh.

Code is based on 3d-point-clouds-HyperCloud, IM-NET-pytorch and IM-NET.

Requirements

• dependencies stored in requirements.txt.
• Python 3.6+
• cuda

Installation

If you are using Conda:

• run bash install_requirements.sh

otherwise:

• install cudatoolkit and run pip install -r requirements.txt

Usage

Add project root directory to PYTHONPATH

export PYTHONPATH=project_path:$PYTHONPATH

Voxels dataset

Download dataset from here and set the dataset location in the hyperparams files in the data_dir key.

Classes can be specified in the hyperparams files in the classes key and the test_classes key

airplane, car, chair, rifle, table

Training

HyperCube

python experiments/train.py --config settings/hyperparams.json

HyperCube-Interval

python experiments/train.py --config settings/hyperparams_interval.json

Latent GAN

python latent_GAN/train.py --input latents.pt --output outdir -c airplane

input is a tensor with latent vectors for class airplane obtained from a trained hypernetwork. c is one of [airplane, car, chair, rifle, table].