face_swap::CNN3DMM Class Reference

This class provided face shape and texture estimation using Caffe with a convolutional neural network. More...

#include <cnn_3dmm.h>

Inheritance diagram for face_swap::CNN3DMM:

Public Member Functions
	CNN3DMM (const std::string &deploy_file, const std::string &caffe_model_file, const std::string &mean_file, bool init_cnn=true, bool with_gpu=true, int gpu_device_id=0)
	Creates an instance of CNN3DMM. More...
	~CNN3DMM ()
	Destructor.
void	process (const cv::Mat &img, cv::Mat &shape_coefficients, cv::Mat &tex_coefficients)
	Estimate face shape and texture from image. More...

Protected Attributes
std::shared_ptr< caffe::Net< float > >	m_net
int	m_num_channels
cv::Size	m_input_size
cv::Mat	m_mean
bool	m_with_gpu

Detailed Description

This class provided face shape and texture estimation using Caffe with a convolutional neural network.

The CNN estimates shape and texture coefficients for a PCA model that is based on Basel's 3D Morphable Model. This is an implementation of the following papers:

1. Regressing Robust and Discriminative 3D Morphable Models with a very Deep Neural Network, Anh Tuan Tran, Tal Hassner, Iacopo Masi and Gerard Medioni.
2. A 3D Face Model for Pose and Illumination Invariant Face Recognition, P. Paysan and R. Knothe and B. Amberg and S. Romdhani and T. Vetter.

Constructor & Destructor Documentation

face_swap::CNN3DMM::CNN3DMM	(	const std::string &	deploy_file,
		const std::string &	caffe_model_file,
		const std::string &	mean_file,
		bool	init_cnn = true,
		bool	with_gpu = true,
		int	gpu_device_id = 0
	)

Creates an instance of CNN3DMM.

Parameters

deploy_file	Path to 3DMM regression CNN deploy file (.prototxt).
caffe_model_file	Path to 3DMM regression CNN model file (.caffemodel).
mean_file	Path to 3DMM regression CNN mean file (.binaryproto).
init_cnn	if true the CNN will be initialized.
with_gpu	Toggle GPU execution.
gpu_device_id	Set the GPU's device id.

Member Function Documentation

void face_swap::CNN3DMM::process	(	const cv::Mat &	img,
		cv::Mat &	shape_coefficients,
		cv::Mat &	tex_coefficients
	)

Estimate face shape and texture from image.

Parameters

[in]	img	The image to process.
[out]	shape_coefficients	PCA shape coefficients.
[out]	tex_coefficients	PCA texture coefficients.

---

The documentation for this class was generated from the following file:

• cnn_3dmm.h