/**
* @brief Main
*
* @param[in] argc
* @param[in] argv
*
* @returns 1
*
*/
int main(int argc, const char *argv[]) {
if (argc < 9) {
banner();
help();
cerr << "ERR: missing parameters" << endl;
return -3;
}
string infile; // = string(argv[1]);
string method(argv[1]);
string config(argv[2]);
string config_e(argv[3]);
cv::Size size(0,0);
int nwidths, nlambdas, nthetas;
size.width = abs(atoi(argv[4]));
size.height = abs(atoi(argv[5]));
nwidths = abs(atoi(argv[6]));
nlambdas= abs(atoi(argv[7]));
nthetas = abs(atoi(argv[8]));
vector<string> classifier_paths;
if (argc >= 10) {
// Read boost XML paths
for (int i = 9; i < argc; i++) {
string clpath(argv[i]);
classifier_paths.push_back(string(argv[i]));
}
} else {
cerr << "ERR: you must specify some classifiers" << endl;
return -2;
}
FacePreProcessor* preprocessor;
EmoDetector* emodetector;
try {
if (config_e == "none") {
preprocessor = new FacePreProcessor(config, size.width, size.height,
nwidths, nlambdas, nthetas);
} else {
double t1 = timestamp();
preprocessor = new FacePreProcessor(config, config_e, size.width,
size.height, nwidths, nlambdas, nthetas);
cout << " FacePreProcessor: " << timestamp() - t1 << " ";
}
if (method == "svm") {
emodetector = new SVMEmoDetector(kCfactor, kMaxIteration, kErrorMargin);
} else {
emodetector = new BoostEmoDetector(kBoostType, kTrimWeight, kMaxDepth);
}
emodetector->init(classifier_paths);
cout << "Insert the image file path: " << endl;
while(std::getline(std::cin, infile)) {
try {
cout << "Processing '" << infile << "'" << endl;
Mat img = matrix_io_load(infile);
Mat features;
bool canPreprocess = preprocessor->preprocess(img, features);
if (!canPreprocess) {
cerr << "ERR: Cannot preprocess this image '" << infile << "'" << endl;
continue;
}
pair<Emotion,float> prediction = emodetector->predict(features);
cout << "Emotion predicted: " << emotionStrings(prediction.first) << " with score " << prediction.second << endl;
cout << "Insert the image file path: " << endl;
} catch (int ee) {
cerr << "ERR: Something wrong with '" << infile << "' (" << ee << ")" << endl;
}
}
delete emodetector;
delete preprocessor;
} catch (int e) {
cerr << "ERR: Exception #" << e << endl;
return -e;
}
return 0;
}
void imageCb(const sensor_msgs::ImageConstPtr& msg)
{
cv_bridge::CvImagePtr cv_ptr;
try
{
cv_ptr = cv_bridge::toCvCopy(msg, sensor_msgs::image_encodings::BGR8);
}
catch (cv_bridge::Exception& e)
{
ROS_ERROR("cv_bridge exception: %s", e.what());
return;
}
Mat img = cv_ptr->image;
Mat features;
cv::cvtColor(img, img, cv::COLOR_BGR2GRAY);
for (int i = 0; i < face_locs.objects.size(); i++)
{
Mat roi = img(cv::Rect(face_locs.objects[i].object.x_offset, face_locs.objects[i].object.y_offset,
face_locs.objects[i].object.width, face_locs.objects[i].object.height)).clone();
bool canPreprocess = preprocessor->preprocess(roi, features);
if (!canPreprocess) {
// cout << " Can't Process" << endl;
}
else {
pair<Emotion, float> prediction = emodetector->predict(features);
emotionString = emotionStrings(prediction.first);
emotionAccuracy = prediction.second;
//emotion_msg.data = emotionStrings(prediction.first);
// ROS_INFO("%s", emotion_msg.data.c_str());
}
//cv::putText(roi, emotionString, textOrg, fontFace, fontScale, CV_RGB(255, 255, 0), thickness, lineType);
// Output modified video stream
// std::string s = SSTR( i );
//cv::imshow(s, roi);
//cv::waitKey(3);
//emotion_pub.publish(emotion_msg);
//image_pub_.publish(cv_ptr->toImageMsg());
cmt_tracker_msgs::Object face_description;
face_description = face_locs.objects[i];
face_description.obj_states.data = emotionString;
face_description.obj_accuracy.data = emotionAccuracy;
emot_pub_faces.objects.push_back(face_description);
}
faces_locations.publish(emot_pub_faces);
face_locs.objects.clear();
emot_pub_faces.objects.clear();
}
