Projects use Mini-Caffe as a C++ library to run CNNs.
Detect facial landmarks with Mini-Caffe. Caffe models are trained by luoyetx/deep-landmark. A video test can be viewed here.
Generate anime face using WGAN. Model is trained by luoyetx/WGAN and converted to Caffe model.
Object Detection via Region-based Fully Convolutional Networks is converted from Orpine/py-R-FCN. Donwload resnet50_rfcn_final.caffemodel then you can run the code.
SSH: Single Stage Headless Face Detector is converted from mahyarnajibi/SSH. Download SSH.caffemodel and run the code on GPU.
SSD: Single Shot MultiBox Detector is converted from chuanqi305/MobileNet-SSD.
You need OpenCV to build examples.
$ mkdir build
$ cd build
$ cmake ..