diff --git a/include/find_object/Settings.h b/include/find_object/Settings.h
index b9559520..d8390469 100644
--- a/include/find_object/Settings.h
+++ b/include/find_object/Settings.h
@@ -189,6 +189,7 @@ class FINDOBJECT_EXP Settings
 	PARAMETER(Feature2D, SIFT_contrastThreshold, double, 0.04, "The contrast threshold used to filter out weak features in semi-uniform (low-contrast) regions. The larger the threshold, the less features are produced by the detector.");
 	PARAMETER(Feature2D, SIFT_edgeThreshold, double, 10, "The threshold used to filter out edge-like features. Note that the its meaning is different from the contrastThreshold, i.e. the larger the edgeThreshold, the less features are filtered out (more features are retained).");
 	PARAMETER(Feature2D, SIFT_sigma, double, 1.6, "The sigma of the Gaussian applied to the input image at the octave #0. If your image is captured with a weak camera with soft lenses, you might want to reduce the number.");
+	PARAMETER(Feature2D, SIFT_rootSIFT, bool, false, "RootSIFT descriptors.");
 
 	PARAMETER(Feature2D, SURF_hessianThreshold, double, 600.0, "Threshold for hessian keypoint detector used in SURF.");
 	PARAMETER(Feature2D, SURF_nOctaves, int, 4, "Number of pyramid octaves the keypoint detector will use.");
diff --git a/src/FindObject.cpp b/src/FindObject.cpp
index 42f048c1..36844400 100644
--- a/src/FindObject.cpp
+++ b/src/FindObject.cpp
@@ -460,6 +460,26 @@ void computeFeatures(
 		}
 		timeExtraction+=timeStep.restart();
 	}
+
+	if( Settings::getFeature2D_SIFT_rootSIFT() &&
+		Settings::currentDescriptorType() == "SIFT" &&
+		!descriptors.empty())
+	{
+		UINFO("Performing RootSIFT...");
+		// see http://www.pyimagesearch.com/2015/04/13/implementing-rootsift-in-python-and-opencv/
+		// apply the Hellinger kernel by first L1-normalizing and taking the
+		// square-root
+		for(int i=0; i<descriptors.rows; ++i)
+		{
+			descriptors.row(i) = descriptors.row(i) / cv::sum(descriptors.row(i))[0];
+			cv::sqrt(descriptors.row(i), descriptors.row(i));
+
+			// By taking the L1 norm, followed by the square-root, we have
+			// already L2 normalized the feature vector and further normalization
+			// is not needed.
+			//descs /= (np.linalg.norm(descs, axis=1, ord=2) + eps);
+		}
+	}
 }