DarkHelp  v1.9.7-1
C++ API for the neural network framework Darknet
Looking for a C++ dev who knows OpenCV?
I'm looking for work. Hire me!
All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Pages
DarkHelp Python API

The DarkHelp Python API uses the simplified DarkHelp C API.

Note
The Python API does not expose 100% of what is available in the DarkHelp C++ API.

Calls such as DarkHelp.EnableNamesIncludePercentage() map to the C API EnableNamesIncludePercentage() which is documented as DarkHelp::Config::names_include_percentage. Using this API, DarkHelp can be used with advanced features such as tiling and snapping.

Example code:

import DarkHelp
dh = DarkHelp.CreateDarkHelpNN("cars.cfg".encode("utf-8"), "cars.names".encode("utf-8"), "cars.weights".encode("utf-8"))
DarkHelp.SetThreshold(dh, 0.35)
DarkHelp.SetAnnotationLineThickness(dh, 1)
DarkHelp.EnableTiles(dh, True)
DarkHelp.EnableCombineTilePredictions(dh, True)
DarkHelp.EnableOnlyCombineSimilarPredictions(dh, False)
DarkHelp.SetTileEdgeFactor(dh, 0.18f)
DarkHelp.SetTileRectFactor(dh, 1.22f)
DarkHelp.PredictFN(dh, "car_01.jpg".encode("utf-8"))
DarkHelp.Annotate(dh, "output.jpg".encode("utf-8"))
json = DarkHelp.GetPredictionResults(dh)
DarkHelp.DestroyDarkHelpNN(dh)
print(json.decode())

See the sample file src-python/example.py which shows how to use the Python API.

Convenient links to some of the API call documentation to get started: