README

Introduction

This repo is the implementation of Tensorformer in Pytorch.

Demo

• Here is an example result of our repo.

Installation

First you have to make sure that you have all dependencies in place. The simplest way to do so, is to use anaconda.

You can create an anaconda environment called mesh_funcspace using

conda env create -f environment.yaml
conda activate tensorformer

Data preparation

• We follow the ONet to generate all the data, including ShapeNet and ABC, where ABC is the guest dataset of ONet.
• After generating the dataset, we can put the data in dataset.

Training

###ShapeNet python main.py --ae --train --phase 1 --iteration 300000 --dataset data/data_per_category/data_per_category/00000000_all/00000000_vox256_img --sample_dir data/output/vessel_64 --sample_vox_size 64

ABC

python main.py --ae --train --phase 1 --iteration 300000 --dataset data/data_per_category/data_per_category/001_ling/001_vox256_img --sample_dir data/output/ling_64 --sample_vox_size 64

Generate Mesh

###ShapeNet python main.py --ae --phase 1 --sample_dir samples/bsp_ae_out --dataset data/data_per_category/data_per_category/00000000_all/00000000_vox256_img --start 0 --end 20 ###ABC python main.py --ae --phase 1 --dataset data/data_per_category/data_per_category/001_ling/001_vox256_img --sample_dir data/output/ling_64 --start 0 --end 100

Evaluation

python evaluate.py