/3D-C2FT

Primary LanguagePythonMIT LicenseMIT

3D-C2FT: Coarse-to-fine Transformer for Multi-view 3D Reconstruction

Introduction

This repository contains the source code for the paper 3D-C2FT: Coarse-to-fine Transformer for Multi-view 3D Reconstruction, which is accepted by ACCV 2022.

Dataset

We use the ShapeNet and Multi-view Real-life datasets, which are available as follows:

Compatibility

We tested the codes with:

  1. PyTorch 1.12.0 with and without GPU under Ubuntu 18.04 and Anaconda3 (Python 3.8 and above)
  2. PyTorch 1.10.2 with and without GPU under Windows 10 and Anaconda3 (Python 3.7 and above)
  3. PyTorch with CPU under MacOS 12.0 (M1) and Anaconda3 (Python 3.7 and above)

Requirements

  1. Anaconda3
  2. PyTorch
  3. Matplotlib
  4. Open3D
  5. PyMCubes
  6. Natsort

Usage

  • Run the code eval.py with the given configuration in config.py
$ python eval.py

Pretrained Model

The pretrained model is available at here:

License

This work is an open-source under MIT license.

Cite this work

@InProceedings{3DC2FT_2022_ACCV,
    author    = {Tiong, Leslie Ching Ow and Sigmund, Dick and Teoh, Andrew Beng Jin},
    title     = {3D-C2FT: Coarse-to-fine Transformer for Multi-view 3D Reconstruction},
    booktitle = {Proceedings of the Asian Conference on Computer Vision (ACCV)},
    month     = {December},
    year      = {2022},
    pages     = {1438-1454}
}