int main(){
FaceRec fr;
fr.train("/home/kevin/workspace/opencv/csv.ext");
vector<int> labels, predictLabels;
predictLabels = fr.predict("/home/kevin/workspace/opencv/test.ext", labels);
for(int i = 0; i < labels.size(); i++){
cout << "Actual result: " << labels[i] << endl;
cout << "Predict result: " << predictLabels[i] << endl;
}
return 0;
}
void executeCB(const corobot_face_recognition::FaceRecognitionGoalConstPtr &goal)
{
if (_as.isPreemptRequested() || !ros::ok()) {
ROS_INFO("%s: Preempted", _action_name.c_str());
// set the action state to preempted
_as.setPreempted();
// success = false;
// break;
}
// helper variables
ros::Rate r(60);
bool success = true;
_goal_id = goal->order_id;
_goal_argument = goal->order_argument;
_feedback.order_id = _goal_id;
_feedback.names.clear();
_feedback.confidence.clear();
_result.order_id = _goal_id;
_result.names.clear();
_result.confidence.clear();
// publish info to the console for the user
ROS_INFO("%s: Executing. order_id: %i, order_argument: %s ", _action_name.c_str(), _goal_id, _goal_argument.c_str());
switch (_goal_id) {
// RECOGNIZE ONCE
case 0:
// cout << "case 0" << endl;
// _fr = new FaceRec(true);
_image_sub = _it.subscribe("/usb_cam/image_raw", 1, &FaceRecognition::recognizeCb, this);
while( _as.isActive() && !_as.isPreemptRequested() && !ros::isShuttingDown() )
r.sleep();
break;
// RECOGNIZE CONTINUOUSLY
case 1:
//cout << "case 1" << endl;
// _fr = new FaceRec(true);
_image_sub = _it.subscribe("/usb_cam/image_raw", 1, &FaceRecognition::recognizeCb, this);
while( _as.isActive() && !_as.isPreemptRequested() && !ros::isShuttingDown() )
r.sleep();
break;
// ADD TRAINING IMAGES
case 2:
// cout << "case 2" << endl;
_fc = new FaceImgCapturer(_goal_argument);
_image_sub = _it.subscribe("/usb_cam/image_raw", 1, &FaceRecognition::addTrainingImagesCb, this);
while( _as.isActive() && !_as.isPreemptRequested() && !ros::isShuttingDown() )
r.sleep();
break;
// TRAIN DATABASE
case 3:
// call training function and check if successful
// bool trainingSuccessful = true;
// cout << "case 3" << endl;
_fr->train(_preprocessing);
if (true)
_as.setSucceeded(_result);
else
_as.setAborted(_result);
break;
// STOP
case 4:
if (_as.isActive()) {
_as.setPreempted();
// _image_sub.shutdown();
}
break;
// EXIT
case 5:
// cout << "case 4" << endl;
ROS_INFO("%s: Exit request.", _action_name.c_str());
_as.setSucceeded(_result);
r.sleep();
ros::shutdown();
break;
}
}
// run face recognition on the recieved image
void recognizeCb(const sensor_msgs::ImageConstPtr& msg) {
// cout << "entering.. recognizeCb" << endl;
_ph.getParam("algorithm", _recognitionAlgo);
if (_as.isPreemptRequested() || !ros::ok()) {
// std::cout << "preempt req or not ok" << std::endl;
ROS_INFO("%s: Preempted", _action_name.c_str());
// set the action state to preempted
_as.setPreempted();
// success = false;
// break;
// cout << "shutting down _image_sub" << endl;
_image_sub.shutdown();
Mat inactive(_window_rows, _window_cols, CV_8UC3, CV_RGB(20,20,20));
appendStatusBar(inactive, "INACTIVE", "");
cv::imshow(_windowName, inactive);
cv::waitKey(3);
return;
}
if (!_as.isActive()) {
// std::cout << "not active" << std::endl;
// cout << "shutting down _image_sub" << endl;
_image_sub.shutdown();
Mat inactive(_window_rows, _window_cols, CV_8UC3, CV_RGB(20,20,20));
appendStatusBar(inactive, "INACTIVE", "");
cv::imshow(_windowName, inactive);
cv::waitKey(3);
return;
}
cv_bridge::CvImagePtr cv_ptr;
try
{
cv_ptr = cv_bridge::toCvCopy(msg, sensor_msgs::image_encodings::BGR8);
}
catch (cv_bridge::Exception& e)
{
ROS_ERROR("%s: cv_bridge exception: %s", _action_name.c_str(), e.what());
return;
}
if (_window_rows == 0) {
_window_rows = cv_ptr->image.rows;
_window_cols = cv_ptr->image.cols;
}
// clear previous feedback
_feedback.names.clear();
_feedback.confidence.clear();
// _result.names.clear();
// _result.confidence.clear();
std::vector<cv::Rect> faces;
std::vector<cv::Mat> faceImgs = _fd.getFaceImgs(cv_ptr->image, faces, true);
std::map<string, std::pair<string, double> > results;
for( size_t i = 0; i < faceImgs.size(); i++ ) {
cv::resize(faceImgs[i], faceImgs[i], cv::Size(100.0, 100.0));
cv::cvtColor( faceImgs[i], faceImgs[i], CV_BGR2GRAY );
cv::equalizeHist( faceImgs[i], faceImgs[i] );
// perform recognition
results = _fr->recognize(faceImgs[i],
("eigenfaces" == _recognitionAlgo),
("fisherfaces" == _recognitionAlgo),
("lbph" == _recognitionAlgo)
);
ROS_INFO("Face %lu:", i);
if ("eigenfaces" == _recognitionAlgo)
ROS_INFO("\tEigenfaces: %s %f", results["eigenfaces"].first.c_str(), results["eigenfaces"].second);
if ("fisherfaces" == _recognitionAlgo)
ROS_INFO("\tFisherfaces: %s %f", results["fisherfaces"].first.c_str(), results["fisherfaces"].second);
if ("lbph" == _recognitionAlgo)
ROS_INFO("\tLBPH: %s %f", results["lbph"].first.c_str(), results["lbph"].second);
}
// update GUI window
// TODO check gui parameter
appendStatusBar(cv_ptr->image, "RECOGNITION", "");
cv::imshow(_windowName, cv_ptr->image);
cv::waitKey(3);
// if faces were detected
if (faceImgs.size() > 0) {
// recognize only once
if (_goal_id == 0) {
// std::cout << "goal_id 0. setting succeeded." << std::endl;
// cout << _recognitionAlgo << endl;
_result.names.push_back(results[_recognitionAlgo].first);
_result.confidence.push_back(results[_recognitionAlgo].second);
_as.setSucceeded(_result);
}
// recognize continuously
else {
_feedback.names.push_back(results[_recognitionAlgo].first);
_feedback.confidence.push_back(results[_recognitionAlgo].second);
_as.publishFeedback(_feedback);
}
}
}
