/A-TVSNet

Code for "A-TVSNet: Aggregated Two-View Stereo Network for Multi-View Stereo Depth Estimation"

Primary LanguagePython

A-TVSNet

Code for "A-TVSNet: Aggregated Two-View Stereo Network for Multi-View Stereo Depth Estimation"

Enviroment

  • Ubuntu 16.04
  • Python 2.7
  • Cuda 9.0 and cudnn 7.0
  • Tensorflow 1.5
  • other dependencies in requirements.txt

Run Examples

  • Extract model.zip to model folder
  • Run example 
    cd atvsnet
python example.py

Reproduce ETH3D Pointcloud Results

  • Download preprocess ETH3D dataset from MVSNet and extract to data folder
  • Build fusible 
    cd fusible
mkdir build
cd build
cmake ..
make
  • Produce point cloud 
    cd atvsnet
./reproduce_pc.sh