dlib_mmod_detect

A dlib mmod object detection sample with a pre-trained eye detector built with the Hunter package manager. This is a simple test bed for experimenting with dlib's dnn toolkit.

Sample test set output for training a small object detector with 500x2=1000 manually annotated eye crops (without occlusion) from unsplash.com face images.

Quick Start

• Install compiler: http://cgold.readthedocs.io/en/latest/first-step.html
• Install CMake (and add to PATH)
• Install Python (for Polly)
• Clone Polly and add <polly>/bin to PATH

export TOOLCHAIN=xcode # for more options: polly.py --help
export CONFIG=Release
polly.py --toolchain ${TOOLCHAIN} --config-all ${CONFIG} --install --verbose --reconfig
find ${HOME}/path/to/images -name "*.jpg" > images.txt
_install/${TOOLCHAIN}/bin/dnn_mmod_find_thing_ex -i images.txt -m _install/xcode/share/dlib_mmod_eye_detector.dat -o /tmp/

...