Installation

Install it from pypi

curl https://sh.rustup.rs -sSf | sh

rustup default nightly

pip install darknetpy

Install a pre-built binary

pip install https://github.com/danielgatis/darknetpy/raw/master/dist/darknetpy-4.1-cp36-cp36m-linux_x86_64.whl

GPU=1 pip install darknetpy

CUDNN=1 pip install darknetpy

OPENCV=1 pip install darknetpy

OPENMP=1 pip install darknetpy

Usage

In example.py:

from darknetpy.detector import Detector

detector = Detector('<absolute-path-to>/darknet/cfg/coco.data',
                    '<absolute-path-to>/darknet/cfg/yolo.cfg',
                    '<absolute-path-to>/darknet/yolo.weights')

results = detector.detect('<absolute-path-to>/darknet/data/dog.jpg')

print(results)

Runing:

python example.py

Result:

[{'right': 194, 'bottom': 353, 'top': 264, 'class': 'dog', 'prob': 0.8198755383491516, 'left': 71}]

Build

On the project root directory

docker pull hoshizora/manylinux1-clang_x86_64

docker run --rm -v `pwd`:/io hoshizora/manylinux1-clang_x86_64 /io/build-wheels.sh