g-apple1/smalify-pig3d

SMALify are optimization-based 3D reconstruction techniques relying on SMAL model. Pig3D as a 3D model can directly replace the SMAL model.

SMALify-Pig3D

SMALify are optimization-based 3D reconstruction techniques relying on SMAL model. Pig3D as a 3D model can directly replace the SMAL model.

How to use

Evironment

We test our codes under python=3.8.15, torch=1.11.0, torchvision=0.12.0, cuda=11.3, pytorch3d=0.7.2. Other versions may not be compatible.

Install dependencies, particularly PyTorch (cuda support recommended), Pytorch3D. Check requirements.txt for full details.

Installation

1. Clone the repository and enter directory

   git clone https://github.com/G-Apple1/SMALify-Pig3D.git
   cd SMALify-Pig3D
   

2. Extract the SMALST.zip into the "/data" directory

3. QuickStart: Run on images in the dataset.

   • 5 images: ['000994', '000906', '000635', '000352', '000372']

     python smal_fitter/optimize_to_joints.py
     

smalify-pig3d's Issues

Does this work relate to a paper?

Hello, I find the fitting results are impressive. Is there any manuscript or paper linked to this project?