torchcpd 1.0.2

PyTorch Implementation of the Coherent Point Drift Algorithm

Torch-CPD

Pytorch Implementation of the Coherent Point Drift Algorithm.

Introduction

This is a PyTorch re-implementation of the excellent pycpd package, which implements the Coherent Point Drift (CPD) algorithm. Only Rigid and Deformable Registration and currently implemented. The default is to use CUDA to speed up computation, which is significantly faster than pure Numpy. Refer to the pycpd package for examples of how to use CPD for registration.

Installation

Install from PyPI

You should install torch separately, follow instructions at https://pytorch.org/get-started/locally/

pip install torchcpd

Installation from Source

git clone https://github.com/ramtingh/torchcpd.git $HOME/torchcpd

Install the package:

pip install .