PattCut Object Detection and Recognition library (http://pattcut.com/)
Main features:
- Small, simple and fast
- Can detect multiple objects at one pass
- Returns extended information about detected areas
- Works with PattCut classifiers and OpenCV cascade classifiers
- Support JPEG, PNG and BMP image formats
Results:
Usage example:
PattCutLib::RecognitionController controller;
auto classifier_id = controller.load_classifier_from_file("classifier.xml");
auto res = controller.perform_recognition("image.jpg");
for (int i = 0; i < res.result_per_classifier(0).areas().size(); i++)
{
auto cur_area = res.result_per_classifier(0).areas(i);
std::cout << "veracity: " << cur_area.veracity() <<
", wnd_count: " << cur_area.merged_window_count() <<
", area: [" << cur_area.rect_x() << ", " << cur_area.rect_y() << "; " << cur_area.rect_width() << ", " << cur_area.rect_height() << "]" << std::endl;
}
Extended usage example:
https://github.com/qoollo/pattcut-odr-library/blob/master/src/RecognitionLibTest/src/main.cpp
Other information: