ngocbh / mtcnn-tf

MTCNN implementation by C++ and tensorflow framework

MTCNN TensorFlow C++ Implementation

Build

Install opencv

• For Unix user

brew install opencv
brew install pkg-config

Tensorflow build

• Flow this !tutorial to build tensorflow c api. A brief:

git clone https://github.com/tensorflow/tensorflow.git
cd tensorflow
bazel build --config=opt //tensorflow/tools/lib_package:libtensorflow

• Copy file bazel-bin/tensorflow/tools/lib_package/libtensorflow.tar.gz to mtcnn-tf directory and unzip it

mkdir -p libs
cd libs
(Copy libtensorflow.tar.gz to this directory)
tar -xvzf libtensorflow.tar.gz

Compile by make

make -j4

Run

Test detect image

./bin/detect_face --input test.jpg --output test-detected.jpg

Test camera

./bin/camera_detect

tensorflow docs

• TF_Session : run a tensor to export output ( the real value )
• TF_Graph : represent dataflow graph
• TF_Status : Denotes success or failure of a call in Tensorflow.

opencv docs

• cv::Mat : image stored
• frame.size() shape of Mat ( cout << frame.size() << endl; )
• cv::imread(fname [,flag]) : read image from file and Flags specifying the color type of a loaded image
• cv::convertTo(img,CV_32FC3) : convert type of Mat to 32-bit floating-point numbers
• cv::cvtColor(working_img,working_img, cv::COLOR_BGR2RGB);