Added face detection

wagtail/wagtailimages/backends/base.py

@@ -58,9 +58,19 @@ class BaseImageBackend(object):
 
     def smart_crop(self, image, size):
         image_mode, image_data = self.image_data_as_rgb(image)
-        focal_points = feature_detection.detect_features(image.size, image_mode, image_data)
 
-        return self.crop_to_points(image, size, focal_points)
+        # Face detection
+        faces = feature_detection.detect_faces(image.size, image_mode, image_data)
+        if faces:
+            return self.crop_to_points(image, size, faces)
+
+        # Feature detection
+        features = feature_detection.detect_features(image.size, image_mode, image_data)
+        if features:
+            return self.crop_to_points(image, size, features)
+
+        # Fall back to crop to centre
+        return self.crop_to_centre(image, size)
 
     def resize_to_max(self, image, size):
         """

wagtail/wagtailimages/utils/feature_detection.py

@@ -1,3 +1,5 @@
+import os
+
 try:
     import cv
 
@@ -14,15 +16,20 @@ except ImportError:
 from wagtail.wagtailimages.utils.focal_point import FocalPoint
 
 
+def get_cv_gray_image(image_size, image_mode, image_data):
+    image = cv.CreateImageHeader(image_size, cv.IPL_DEPTH_8U, 3)
+    cv.SetData(image, image_data)
+
+    gray_image = cv.CreateImage(image_size, 8, 1)
+    convert_mode = getattr(cv, 'CV_%s2GRAY' % image_mode)
+    cv.CvtColor(image, gray_image, convert_mode)
+
+    return gray_image
+
+
 def detect_features(image_size, image_mode, image_data):
     if opencv_available:
-        image = cv.CreateImageHeader(image_size, cv.IPL_DEPTH_8U, 3)
-        cv.SetData(image, image_data)
-
-        gray_image = cv.CreateImage(image_size, 8, 1)
-        convert_mode = getattr(cv, 'CV_%s2GRAY' % image_mode)
-        cv.CvtColor(image, gray_image, convert_mode)
-        image = gray_image
+        image = get_cv_gray_image(image_size, image_mode, image_data)
         rows = image_size[0]
         cols = image_size[1]
 
@@ -34,3 +41,27 @@ def detect_features(image_size, image_mode, image_data):
             return [FocalPoint(x, y, 1) for x, y in points]
 
     return []
+
+
+def detect_faces(image_size, image_mode, image_data):
+    if opencv_available:
+        cascade_filename = os.path.join(os.path.dirname(__file__), 'face_detection', 'haarcascade_frontalface_alt.xml')
+        cascade = cv.Load(cascade_filename)
+        image = get_cv_gray_image(image_size, image_mode, image_data)
+
+        cv.EqualizeHist(image, image)
+
+        min_size = (40, 40)
+        haar_scale = 1.1
+        min_neighbors = 3
+        haar_flags = 0
+
+        faces = cv.HaarDetectObjects(
+            image, cascade, cv.CreateMemStorage(0),
+            haar_scale, min_neighbors, haar_flags, min_size
+        )
+
+        if faces:
+            return [FocalPoint.from_square(face[0][0], face[0][1], face[0][2], face[0][3]) for face in faces]
+
+    return []