diff --git a/face.go b/face.go
index 8fe2fc6..b59010a 100644
--- a/face.go
+++ b/face.go
@@ -16,9 +16,9 @@ import (
 )
 
 const (
-	shapeLen = 2
 	rectLen  = 4
 	descrLen = 128
+	shapeLen = 2
 )
 
 // A Recognizer creates face descriptors for provided images and
@@ -90,7 +90,6 @@ func (rec *Recognizer) recognize(type_ int, imgData []byte, maxFaces int) (faces
 	cImgData := (*C.uint8_t)(&imgData[0])
 	cLen := C.int(len(imgData))
 	cMaxFaces := C.int(maxFaces)
-
 	cType := C.int(type_)
 
 	ret := C.facerec_recognize(rec.ptr, cImgData, cLen, cMaxFaces, cType)
@@ -102,13 +101,14 @@ func (rec *Recognizer) recognize(type_ int, imgData []byte, maxFaces int) (faces
 		return
 	}
 
-	// No faces.
 	numFaces := int(ret.num_faces)
 	if numFaces == 0 {
 		return
 	}
+	numShapes := int(ret.num_shapes)
 
 	// Copy faces data to Go structure.
+	defer C.free(unsafe.Pointer(ret.shapes))
 	defer C.free(unsafe.Pointer(ret.rectangles))
 	defer C.free(unsafe.Pointer(ret.descriptors))
 
@@ -120,6 +120,10 @@ func (rec *Recognizer) recognize(type_ int, imgData []byte, maxFaces int) (faces
 	dDataPtr := unsafe.Pointer(ret.descriptors)
 	dData := (*[1 << 30]float32)(dDataPtr)[:dDataLen:dDataLen]
 
+	sDataLen := numFaces * numShapes * shapeLen
+	sDataPtr := unsafe.Pointer(ret.shapes)
+	sData := (*[1 << 30]C.long)(sDataPtr)[:sDataLen:sDataLen]
+
 	for i := 0; i < numFaces; i++ {
 		face := Face{}
 		x0 := int(rData[i*rectLen])
@@ -128,6 +132,11 @@ func (rec *Recognizer) recognize(type_ int, imgData []byte, maxFaces int) (faces
 		y1 := int(rData[i*rectLen+3])
 		face.Rectangle = image.Rect(x0, y0, x1, y1)
 		copy(face.Descriptor[:], dData[i*descrLen:(i+1)*descrLen])
+		for j := 0; j < numShapes; j++ {
+			shapeX := int(sData[(i*numShapes+j)*shapeLen])
+			shapeY := int(sData[(i*numShapes+j)*shapeLen+1])
+			face.Shapes = append(face.Shapes, image.Point{shapeX, shapeY})
+		}
 		faces = append(faces, face)
 	}
 	return
diff --git a/facerec.cc b/facerec.cc
index 9dac433..cd09c54 100644
--- a/facerec.cc
+++ b/facerec.cc
@@ -45,9 +45,9 @@ using anet_type = loss_metric<fc_no_bias<128,avg_pool_everything<
                             input_rgb_image_sized<150>
                             >>>>>>>>>>>>;
 
-static const size_t SHAPE_LEN = 2;
 static const size_t RECT_LEN = 4;
 static const size_t DESCR_LEN = 128;
+static const size_t SHAPE_LEN = 2;
 static const size_t RECT_SIZE = RECT_LEN * sizeof(long);
 static const size_t DESCR_SIZE = DESCR_LEN * sizeof(float);
 static const size_t SHAPE_SIZE = SHAPE_LEN * sizeof(long);
@@ -209,13 +209,13 @@ faceret* facerec_recognize(facerec* rec, const uint8_t* img_data, int len, int m
 	}
 	ret->num_shapes = shapes[0].num_parts();
 	ret->shapes = (long*)malloc(ret->num_faces * ret->num_shapes * SHAPE_SIZE);
-	for (int i = 0; i < 	ret->num_faces; i++) {
+	for (int i = 0; i < ret->num_faces; i++) {
 		long* dst = ret->shapes + i * ret->num_shapes * SHAPE_LEN;
-		auto shape = shapes[i];
-		for (long unsigned int j = 0;j < shape.num_parts(); j++) {
-			dst[j * SHAPE_LEN] = shape.part(j).x();
-			dst[(j * SHAPE_LEN) + 1] = shape.part(j).y();
-        }
+		const auto& shape = shapes[i];
+		for (int j = 0; j < ret->num_shapes; j++) {
+			dst[j*SHAPE_LEN] = shape.part(j).x();
+			dst[j*SHAPE_LEN+1] = shape.part(j).y();
+		}
 	}
 	return ret;
 }