Authors: [Shi-Sheng Huang], please contact with me via shishenghuang.net@gmail.com
This is a C++ Inteface of MinkowskiNet, tested in Ubuntu 16.04,The details can be found in the following CVPR 2019 paper
Christopher B. Choy, JunYoung Gwak, Silvio Savarese: 4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks. CVPR 2019: 3075-3084
We have tested the library in 16.04, but it should be easy to compile in other platforms. A powerful computer (e.g. i7) will ensure real-time performance and provide more stable and accurate results.
We use the new thread and chrono functionalities of C++11.
REQUIRED for all GPU accelerated code at least with cmake it is still possible to compile the CPU part without available at https://developer.nvidia.com/cuda-downloads
If you have installed CUDA 10.1, just change your cuda to CUDA 10.1 if more than one CUDA version is installed
sudo apt-get install libgflags-dev libgoogle-glog-dev
sudo add-apt-repository ppa:deadsnakes/ppa
sudo apt update
sudo apt install python3.6
sudo apt-get install python3.6-dev
./python_env.sh
then you have install python3.6 virtual environment.
wget https://node1.chrischoy.org/data/publications/minknet/MinkUNet34C-train-conv1-5.pth
just set MinkUNet34C-train-conv1-5.pth path located in your computer.
mkdir build
cd build
cmake ..
make -j8
cd bin
./test
If any problem, please check Spatio-Temporal Segmentation