dlib extracting chip from object detector

Question

I'm using dlib for object detection, labeling and training dataset is completed using HOG+SVM, it is able to detect the object, but now

1. I need extract chip from the image

2. I would like to add label to detected area on win_overlay. my working code snippet here, i searched for both of this, but I could not find for object detection.

   typedef scan_fhog_pyramid<pyramid_down<6> > image_scanner_type;
   image_scanner_type scanner;
   
   object_detector<image_scanner_type> detector;
   deserialize(argv[2]) >> detector;
   array2d<unsigned char> img;
   image_window win;
   load_image(img, "test.jpg");
   std::vector<rectangle> dets = detector(img);
   if (dets.size() > 0) {
   win.clear_overlay();
   win.set_image(img);
   win.add_overlay(dets, rgb_pixel(255, 0, 0));
   }
   

   above snippet is highlighting detection, but i would like to label detected are on "win.add_overlay()" and extract chip(detected area) and save as image.

Answer 1

1. You should take a look on this example : http://dlib.net/face_landmark_detection_ex.cpp.html

dlib function extract_image_chips is the way to extract chip from img

2. In this case: i usually use OpenCV to do this. After you get rectangle from dlib object detector. Convert it to OpenCV cv::Rect . Create a text and imshow all.