|
|
|
|
@ -6,13 +6,13 @@
|
|
|
|
|
namespace easypr {
|
|
|
|
|
|
|
|
|
|
Svm::Svm(const char* forward_data_folder, const char* inverse_data_folder)
|
|
|
|
|
: forward_(forward_data_folder), inverse_(inverse_data_folder) {
|
|
|
|
|
: forward_(forward_data_folder), inverse_(inverse_data_folder) {
|
|
|
|
|
assert(forward_);
|
|
|
|
|
assert(inverse_);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void Svm::divide(const char* images_folder, float percentage /* = 0.7 */) {
|
|
|
|
|
auto files = Utils::getFiles(images_folder);
|
|
|
|
|
auto files = utils::getFiles(images_folder);
|
|
|
|
|
if (files.empty()) {
|
|
|
|
|
std::cout << "No file found in " << images_folder << std::endl;
|
|
|
|
|
return;
|
|
|
|
|
@ -25,14 +25,14 @@ void Svm::divide(const char* images_folder, float percentage /* = 0.7 */) {
|
|
|
|
|
for (size_t i = 0; i < files.size(); ++i) {
|
|
|
|
|
// TODO: Move files directly to improve efficiency.
|
|
|
|
|
auto f = files[i];
|
|
|
|
|
auto file_name = Utils::getFileName(f, true).c_str();
|
|
|
|
|
auto file_name = utils::getFileName(f, true);
|
|
|
|
|
auto image = cv::imread(f);
|
|
|
|
|
char save_to[255] = {0};
|
|
|
|
|
assert(!image.empty());
|
|
|
|
|
if (i < split_index) {
|
|
|
|
|
sprintf(save_to, "%s/train/%s", images_folder, file_name);
|
|
|
|
|
sprintf(save_to, "%s/train/%s", images_folder, file_name.c_str());
|
|
|
|
|
} else {
|
|
|
|
|
sprintf(save_to, "%s/test/%s", images_folder, file_name);
|
|
|
|
|
sprintf(save_to, "%s/test/%s", images_folder, file_name.c_str());
|
|
|
|
|
}
|
|
|
|
|
utils::imwrite(save_to, image);
|
|
|
|
|
std::cout << f << " -> " << save_to << std::endl;
|
|
|
|
|
@ -62,7 +62,7 @@ void Svm::get_train() {
|
|
|
|
|
features = features.reshape(1, 1);
|
|
|
|
|
|
|
|
|
|
this->trainingData_.push_back(features);
|
|
|
|
|
labels.push_back(Label::kForward); // Note here
|
|
|
|
|
labels.push_back(Label::kForward); // Note here
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// iterate inverse data
|
|
|
|
|
@ -84,7 +84,7 @@ void Svm::get_train() {
|
|
|
|
|
features = features.reshape(1, 1);
|
|
|
|
|
|
|
|
|
|
this->trainingData_.push_back(features);
|
|
|
|
|
labels.push_back(Label::kInverse); // Note here
|
|
|
|
|
labels.push_back(Label::kInverse); // Note here
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
//
|
|
|
|
|
@ -154,8 +154,10 @@ void Svm::train(bool divide /* = true */, float divide_percentage /* = 0.7 */,
|
|
|
|
|
// need to be trained first
|
|
|
|
|
CvSVMParams SVM_params;
|
|
|
|
|
SVM_params.svm_type = CvSVM::C_SVC;
|
|
|
|
|
//SVM_params.kernel_type = CvSVM::LINEAR; //CvSVM::LINEAR; 线型，也就是无核
|
|
|
|
|
SVM_params.kernel_type = CvSVM::RBF; //CvSVM::RBF 径向基函数，也就是高斯核
|
|
|
|
|
// SVM_params.kernel_type = CvSVM::LINEAR; //CvSVM::LINEAR;
|
|
|
|
|
// 线型，也就是无核
|
|
|
|
|
SVM_params.kernel_type =
|
|
|
|
|
CvSVM::RBF; // CvSVM::RBF 径向基函数，也就是高斯核
|
|
|
|
|
SVM_params.degree = 0.1;
|
|
|
|
|
SVM_params.gamma = 1;
|
|
|
|
|
SVM_params.coef0 = 0.1;
|
|
|
|
|
@ -167,27 +169,25 @@ void Svm::train(bool divide /* = true */, float divide_percentage /* = 0.7 */,
|
|
|
|
|
std::cout << "Generating svm model file, please wait..." << std::endl;
|
|
|
|
|
|
|
|
|
|
try {
|
|
|
|
|
//CvSVM svm(trainingData, classes, cv::Mat(), cv::Mat(), SVM_params);
|
|
|
|
|
// CvSVM svm(trainingData, classes, cv::Mat(), cv::Mat(), SVM_params);
|
|
|
|
|
svm.train_auto(this->trainingData_, this->classes_, cv::Mat(),
|
|
|
|
|
cv::Mat(),
|
|
|
|
|
SVM_params,
|
|
|
|
|
10,
|
|
|
|
|
cv::Mat(), SVM_params, 10,
|
|
|
|
|
CvSVM::get_default_grid(CvSVM::C),
|
|
|
|
|
CvSVM::get_default_grid(CvSVM::GAMMA),
|
|
|
|
|
CvSVM::get_default_grid(CvSVM::P),
|
|
|
|
|
CvSVM::get_default_grid(CvSVM::NU),
|
|
|
|
|
CvSVM::get_default_grid(CvSVM::COEF),
|
|
|
|
|
CvSVM::get_default_grid(CvSVM::DEGREE),
|
|
|
|
|
true);
|
|
|
|
|
CvSVM::get_default_grid(CvSVM::DEGREE), true);
|
|
|
|
|
} catch (const cv::Exception& err) {
|
|
|
|
|
std::cout << err.what() << std::endl;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
utils::mkdir(out_svm_path);
|
|
|
|
|
cv::FileStorage fsTo(out_svm_path, cv::FileStorage::WRITE);
|
|
|
|
|
svm.write(*fsTo, "svm");
|
|
|
|
|
|
|
|
|
|
std::cout << "Generate done! The model file is located at " <<
|
|
|
|
|
out_svm_path << std::endl;
|
|
|
|
|
std::cout << "Generate done! The model file is located at "
|
|
|
|
|
<< out_svm_path << std::endl;
|
|
|
|
|
} else {
|
|
|
|
|
// don't train, use ready-made model file
|
|
|
|
|
try {
|
|
|
|
|
@ -196,10 +196,10 @@ void Svm::train(bool divide /* = true */, float divide_percentage /* = 0.7 */,
|
|
|
|
|
std::cout << err.what() << std::endl;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
} // if train
|
|
|
|
|
} // if train
|
|
|
|
|
|
|
|
|
|
// TODO Check whether the model file exists or not.
|
|
|
|
|
svm.load(out_svm_path, "svm"); // make sure svm model was loaded
|
|
|
|
|
svm.load(out_svm_path, "svm"); // make sure svm model was loaded
|
|
|
|
|
|
|
|
|
|
// 30% testing procedure
|
|
|
|
|
this->get_test();
|
|
|
|
|
@ -220,17 +220,13 @@ void Svm::train(bool divide /* = true */, float divide_percentage /* = 0.7 */,
|
|
|
|
|
cv::Mat out;
|
|
|
|
|
features.convertTo(out, CV_32FC1);
|
|
|
|
|
|
|
|
|
|
Label predict = ((int) svm.predict(out)) == 1 ? kForward : kInverse;
|
|
|
|
|
Label predict = ((int)svm.predict(out)) == 1 ? kForward : kInverse;
|
|
|
|
|
Label real = test_labels_[label_index++];
|
|
|
|
|
|
|
|
|
|
if (predict == kForward && real == kForward)
|
|
|
|
|
ptrue_rtrue++;
|
|
|
|
|
if (predict == kForward && real == kInverse)
|
|
|
|
|
ptrue_rfalse++;
|
|
|
|
|
if (predict == kInverse && real == kForward)
|
|
|
|
|
pfalse_rtrue++;
|
|
|
|
|
if (predict == kInverse && real == kInverse)
|
|
|
|
|
pfalse_rfalse++;
|
|
|
|
|
if (predict == kForward && real == kForward) ptrue_rtrue++;
|
|
|
|
|
if (predict == kForward && real == kInverse) ptrue_rfalse++;
|
|
|
|
|
if (predict == kInverse && real == kForward) pfalse_rtrue++;
|
|
|
|
|
if (predict == kInverse && real == kInverse) pfalse_rfalse++;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
std::cout << "count_all: " << count_all << std::endl;
|
|
|
|
|
@ -244,7 +240,8 @@ void Svm::train(bool divide /* = true */, float divide_percentage /* = 0.7 */,
|
|
|
|
|
precise = ptrue_rtrue / (ptrue_rtrue + ptrue_rfalse);
|
|
|
|
|
std::cout << "precise: " << precise << std::endl;
|
|
|
|
|
} else {
|
|
|
|
|
std::cout << "precise: " << "NA" << std::endl;
|
|
|
|
|
std::cout << "precise: "
|
|
|
|
|
<< "NA" << std::endl;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
double recall = 0;
|
|
|
|
|
@ -252,7 +249,8 @@ void Svm::train(bool divide /* = true */, float divide_percentage /* = 0.7 */,
|
|
|
|
|
recall = ptrue_rtrue / (ptrue_rtrue + pfalse_rtrue);
|
|
|
|
|
std::cout << "recall: " << recall << std::endl;
|
|
|
|
|
} else {
|
|
|
|
|
std::cout << "recall: " << "NA" << std::endl;
|
|
|
|
|
std::cout << "recall: "
|
|
|
|
|
<< "NA" << std::endl;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
double Fsocre = 0;
|
|
|
|
|
@ -260,9 +258,9 @@ void Svm::train(bool divide /* = true */, float divide_percentage /* = 0.7 */,
|
|
|
|
|
Fsocre = 2 * (precise * recall) / (precise + recall);
|
|
|
|
|
std::cout << "Fsocre: " << Fsocre << std::endl;
|
|
|
|
|
} else {
|
|
|
|
|
std::cout << "Fsocre: " << "NA" << std::endl;
|
|
|
|
|
std::cout << "Fsocre: "
|
|
|
|
|
<< "NA" << std::endl;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
} // namespace easypr
|
|
|
|
|
} // namespace easypr
|
|
|
|
|
|
Micooz
10 years ago