add variation of predict that also returns the confidence

ext/opencv/facerecognizer.cpp

@@ -106,6 +106,30 @@ rb_predict(VALUE self, VALUE src)
   return INT2NUM(label);
 }
 
+/*
+ * call-seq:
+ *   predict_with_confidence(src)
+ * 
+ * Predicts a label and associated confidence (e.g. distance) for a given input image.
+ */
+VALUE
+rb_predict_with_confidence(VALUE self, VALUE src)
+{
+  cv::Mat mat = cv::Mat(CVMAT_WITH_CHECK(src));
+  cv::FaceRecognizer *self_ptr = FACERECOGNIZER(self);
+  int label;
+  double confidence;
+  try {
+    self_ptr->predict(mat, label, confidence);
+  }
+  catch (cv::Exception& e) {
+    raise_cverror(e);
+  }
+
+  return rb_ary_new3(2, INT2NUM(label), DBL2NUM(confidence));
+}
+
+
 /*
  * call-seq:
  *   save(filename)
@@ -164,6 +188,7 @@ define_ruby_class()
   rb_klass = rb_define_class_under(opencv, "FaceRecognizer", cAlgorithm::rb_class());
   rb_define_method(rb_klass, "train", RUBY_METHOD_FUNC(rb_train), 2);
   rb_define_method(rb_klass, "predict", RUBY_METHOD_FUNC(rb_predict), 1);
+  rb_define_method(rb_klass, "predict_with_confidence", RUBY_METHOD_FUNC(rb_predict_with_confidence), 1);
   rb_define_method(rb_klass, "save", RUBY_METHOD_FUNC(rb_save), 1);
   rb_define_method(rb_klass, "load", RUBY_METHOD_FUNC(rb_load), 1);
 }

test/test_eigenfaces.rb

@@ -55,6 +55,19 @@ class TestEigenFaces < OpenCVTestCase
     }
   end
 
+  def test_predict_with_confidence
+    img = CvMat.load(FILENAME_LENA256x256, CV_LOAD_IMAGE_GRAYSCALE)
+    label = 1
+    @eigenfaces.train([img], [label])
+    lbl, conf = @eigenfaces.predict_with_confidence(img)
+    assert_equal(label, lbl)
+    assert_equal(0.0, conf)
+
+    assert_raise(TypeError) {
+      @eigenfaces.predict_with_confidence(DUMMY_OBJ)
+    }
+  end
+
   def test_save
     img = CvMat.load(FILENAME_LENA256x256, CV_LOAD_IMAGE_GRAYSCALE)
     label = 1

test/test_fisherfaces.rb

@@ -52,6 +52,17 @@ class TestFisherFaces < OpenCVTestCase
     }
   end
 
+  def test_predict_with_confidence
+    label = 1
+    lbl, conf = @fisherfaces_trained.predict_with_confidence(@images[0])
+    assert_equal(1, lbl)
+    assert_equal(0.0, conf)
+
+    assert_raise(TypeError) {
+      @fisherfaces_trained.predict_with_confidence(DUMMY_OBJ)
+    }
+  end
+
   def test_save
     filename = "fisherfaces_save-#{DateTime.now.strftime('%Y%m%d%H%M%S')}.xml"
     begin