This is a (Cython-based) Python wrapper for Philipp Krähenbühl's Fully-Connected CRFs (version 2).
If you use this code for your reasearch, please cite:
Efficient Inference in Fully Connected CRFs with Gaussian Edge Potentials
Philipp Krähenbühl and Vladlen Koltun
NIPS 2011
- Win64 python2.7
- Win64 python3.5
- Win64 MATLAB2016 <denseCRF_matlab>
- Download code
git clone https://github.com/liyemei/densecrf
- Renameing
Renameing the file "pydensecrf-py27" or
"pydensecrf-py35" to the new name "pydensecrf"
- Installing package
Puting this file "pydensecrf" in this directory "C:\Program Files\Anaconda3\Lib\site-packages"
- denseCRF_matlab can be run directly
- Python
Run on the command line
cd examples/
python inference.py im1.png anno1.png out.png
- Matlab
run the demo.m
https://github.com/ClaireXie/denseCRF_matlab
https://github.com/lucasb-eyer/pydensecrf