VNLB.md

VNLB

This file describes how to create the output files to compate the C++ Code with this Python API. First, the original C++ code must be installed separately from this link.

$ git clone https://github.com/pariasm/vnlb
$ mkdir build; cd build
$ cmake ..
$ make
$ cd ..

Next we show how to use this code to create the ground-truth images. We first move to the directory with the executable files,

$ cd ./build/bin/

Then we run the C++ Code. This particular execution uses frames from 0 to 4 for a total of 5 frames. The noise level is 20.

$ export OMP_NUM_THREADS=4
$ ./vnlb-gt.sh $PYVNLB_HOME/data/davis_baseball_64x64/%05d.jpg 0 4 20 $PYVNLB_HOME/data/davis_baseball_64x64/vnlb/ "-px1 7 -pt1 2 -px2 7 -pt2 2 -verbose"

Finally, write the noise level to a text file in the output directory,

$ echo "20" > $PYVNLB_HOME/pyvnlb/data/davis_baseball_64x64/vnlb/sigma.txt