MTCNN Joint Face Detection and Alignment using Multi-task Cascaded Convolutional Networks.

This project is forked from: https://github.com/wowo200/MTCNN . Which itself is a C++ version of: https://github.com/kpzhang93/MTCNN_face_detection_alignment

The models used here are trained and copied from the original project (https://github.com/kpzhang93/MTCNN_face_detection_alignment).

This fork fixes some bugs over the original C++ version. Moreover, the source code layout is changed so that it can be used as 3rd party from other libraries.

• OpenCV (tested with 3.4)
• Caffe (built with CUDA -- tested )
• Boost (tested with 1.66)

$ cd projects
$ git clone https://github.com/vvirag/MTCNN.git
$ cd MTCNN
$ cmake-gui

After selecting source and build directories, click Configure. If Caffe not found, define the path manually by setting up Caffe_INCLUDE_DIRS and Caffe_LIBS. Afterwards click Generate.

$ cd build
$ make

$ cd projects/MTCNN
$ ./build/example/main.bin ../path_to_image_folder_containing_jpgs_or_jpegs/