Skip to content

Fast, accurate and easy to run dense optical flow with python wrapper

License

Notifications You must be signed in to change notification settings

philferriere/pyflow

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

16 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

This fork modifies Deepak Pathak's (@pathak22)original implementation so that it can properly build on Windows. Based on the OS, setup.py sets the proper compiler flags. If setting up on Windows, make sure the Visual C++ build tools are on your path (see below).

This fork also picks up changes made by Michael Williams (@mpwillia). See here for details.

Python Dense Optical Flow

This is a Python wrapper for Ce Liu's C++ implementation of Coarse2Fine Optical Flow. It's super fast and accurate optical flow method based on Coarse2Fine warping method from Thomas Brox. This python wrapper has minimal dependencies, and it also eliminates the need for C++ OpenCV library. For real time performance, one can additionally resize the images to a smaller size.

For more information about Deepak's wrapper, see here.

Setup instructions (all platforms)

First, make sure the cython package has been installed as part of your virtualenv or conda environment.
Then, to install this version of pyflow, use pip as follows:

(dlwin36tfvos) pferr@MSI E:\repos                                                                                 
$ pip install git+https://github.com/philferriere/pyflow.git                                                      
Collecting git+https://github.com/philferriere/pyflow.git                                                         
  Cloning https://github.com/philferriere/pyflow.git to c:\users\pferr\appdata\local\temp\pip-amvdc0ec-build      
Installing collected packages: pyflow                                                                             
  Running setup.py install for pyflow ... done                                                                    
Successfully installed pyflow-1.0                                                                                 

Note: If distutils uses a different compiler than cl.exe, modify its configuration file as follows:

(dlwin36tfvos) pferr@MSI E:\repos
$ cat %CONDA_PREFIX%\Lib\distutils\distutils.cfg
[build]
compiler=msvc

Additional setup instructions (Windows 10)

For this package to build, you must have the Visual C++ 2015 build tools on your path. If you don't, install them from here:

Then, run visualcppbuildtools_full.exe and select default options:

About

Fast, accurate and easy to run dense optical flow with python wrapper

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Jupyter Notebook 95.5%
  • C++ 4.4%
  • Other 0.1%