Adds the ability to choose between the Fisher and LBPH recognizer

SimpleCV/Features/FaceRecognizer.py

@@ -4,7 +4,7 @@
 
 class FaceRecognizer():
 
-    def __init__(self):
+    def __init__(self, algo="fisher"):
         """
         Create a Face Recognizer Class using Fisher Face Recognizer. Uses
         OpenCV's FaceRecognizer class. Currently supports Fisher Faces.
@@ -19,14 +19,21 @@ def __init__(self):
         self.labels_set = []
         self.int_labels = []
         self.labels_dict_rev = {}
+        self.algo = algo
         # Not yet supported
         # self.eigenValues = None
         # self.eigenVectors = None
         # self.mean = None
 
         try:
             import cv2
-            self.model = cv2.createFisherFaceRecognizer()
+            if self.algo == "fisher":
+                self.model = cv2.createFisherFaceRecognizer()
+            elif self.algo == "lbph":
+                self.model = cv2.createLBPHFaceRecognizer()
+            else:
+                raise ImportError
+
         except ImportError, AttributeError:
             self.supported = False
             warnings.warn("Fisher Recognizer is supported by OpenCV >= 2.4.4")

SimpleCV/tests/tests.py

@@ -3776,7 +3776,7 @@ def test_motionBlur():
     else:
         assert False
 
-def test_faceRecognize():
+def test_fisherfaceRecognize():
     try:
         import cv2
         if hasattr(cv2, "createFisherFaceRecognizer"):
@@ -3832,6 +3832,62 @@ def test_faceRecognize():
     except ImportError:
         pass
 
+def test_lbphfaceRecognize():
+    try:
+        import cv2
+        if hasattr(cv2, "createLBPHFaceRecognizer"):
+            f = FaceRecognizer(algo="lbph")
+            images1 = ["../sampleimages/ff1.jpg",
+                       "../sampleimages/ff2.jpg",
+                       "../sampleimages/ff3.jpg",
+                       "../sampleimages/ff4.jpg",
+                        "../sampleimages/ff5.jpg"]
+
+            images2 = ["../sampleimages/fm1.jpg",
+                       "../sampleimages/fm2.jpg",
+                       "../sampleimages/fm3.jpg",
+                       "../sampleimages/fm4.jpg",
+                       "../sampleimages/fm5.jpg"]
+
+            images3 = ["../sampleimages/fi1.jpg",
+                       "../sampleimages/fi2.jpg",
+                       "../sampleimages/fi3.jpg",
+                       "../sampleimages/fi4.jpg"]
+
+            imgset1 = []
+            imgset2 = []
+            imgset3 = []
+
+            for img in images1:
+                imgset1.append(Image(img))
+            label1 = ["female"]*len(imgset1)
+
+            for img in images2:
+                imgset2.append(Image(img))
+            label2 = ["male"]*len(imgset2)
+
+            imgset = imgset1 + imgset2
+            labels = label1 + label2
+            imgset[4] = imgset[4].resize(400,400)
+            f.train(imgset, labels)
+
+            for img in images3:
+                imgset3.append(Image(img))
+            imgset[2].resize(300, 300)
+            label = []
+            for img in imgset3:
+                name, confidence = f.predict(img)
+                label.append(name)
+
+            if label == ["female", "male", "female", "male"]:
+                pass
+            else:
+                assert False
+        else:
+            pass
+    except ImportError:
+        pass
+
 def test_channelMixer():
     i = Image('lenna')
     r = i.channelMixer()