C++ implementation of Open Oceans
See https://github.com/jonm3D/OpenOceans
dnf -y install \
gcc \
cmake \
cppcheck \
python3.12 \
parallel \
gmp-devel
$ make build test
Create symlink to the input data:
$ ln -s <path to data>/data
$ make classify
...
Classifed CSV files will be written to a predictions
directory. Each
input granule CSV will have a corresponding classified .csv
file and
a .txt
file containing performance statistics.
$ ls -1 predictions/
ATL03_20230213042035_08341807_006_01_gt2l_0_classified.csv
ATL03_20230328030558_01031906_006_01_gt2l_0_classified.csv
ATL03_20230407113920_02611908_006_01_gt1l_0_classified.csv
ATL03_20230407113920_02611908_006_01_gt2l_0_classified.csv
ATL03_20230407113920_02611908_006_01_gt3l_0_classified.csv
...
$ make score
Reading filenames from stdin
180 filenames read
...
cls acc F1 bal_acc cal_F1 tp tn fp fn support total
0 0.983 0.867 0.924 0.914 26346 448235 3638 4464 30810 482683
40 0.998 0.952 0.981 0.980 11044 470520 696 423 11467 482683
41 0.985 0.992 0.951 0.913 437142 38460 3817 3264 440406 482683
...
Surface
Average Acc = 0.944
Average F1 = 0.954
Average BA = 0.925
...
Bathy
Average Acc = 0.980
Average F1 = 0.603
Average BA = 0.893