Merge pull request #346 from blueprintparadise/master

Add MediapipeWrapper as a detector

deepface/detectors/FaceDetector.py

@@ -1,4 +1,4 @@
-from deepface.detectors import OpenCvWrapper, SsdWrapper, DlibWrapper, MtcnnWrapper, RetinaFaceWrapper
+from deepface.detectors import OpenCvWrapper, SsdWrapper, DlibWrapper, MtcnnWrapper, RetinaFaceWrapper,MediapipeWrapper
 from PIL import Image
 import math
 import numpy as np
@@ -13,7 +13,8 @@ def build_model(detector_backend):
         'ssd': SsdWrapper.build_model,
         'dlib': DlibWrapper.build_model,
         'mtcnn': MtcnnWrapper.build_model,
-        'retinaface': RetinaFaceWrapper.build_model
+        'retinaface': RetinaFaceWrapper.build_model,
+	'mediapipe': MediapipeWrapper.build_model
     }
 
     if not "face_detector_obj" in globals():
@@ -50,7 +51,8 @@ def detect_faces(face_detector, detector_backend, img, align = True):
         'ssd': SsdWrapper.detect_face,
         'dlib': DlibWrapper.detect_face,
         'mtcnn': MtcnnWrapper.detect_face,
-        'retinaface': RetinaFaceWrapper.detect_face
+        'retinaface': RetinaFaceWrapper.detect_face,
+	'mediapipe': MediapipeWrapper.detect_face
     }
 
     detect_face = backends.get(detector_backend)

deepface/detectors/MediapipeWrapper.py

@@ -0,0 +1,77 @@
+
+from deepface.detectors import FaceDetector
+
+
+# Link - https://google.github.io/mediapipe/solutions/face_detection
+
+def build_model():
+    import mediapipe as mp
+    mp_face_detection = mp.solutions.face_detection
+    # Build a face detector
+    # min_detection_confidence - "A filter to analyse the training photographs"
+    face_detection =  mp_face_detection.FaceDetection( min_detection_confidence=0.6)
+    return face_detection
+
+def detect_face(face_detector, img, align=True):
+    import mediapipe as mp
+    import re
+    #mp_face_detection = mp.solutions.face_detection
+    #mp_drawing = mp.solutions.drawing_utils
+    resp = []
+    results = face_detector.process(img)
+    original_size = img.shape
+    target_size = (300, 300)
+    # First face , than eye
+    #print(results.detections)
+    if  results.detections:
+        for detection in results.detections:
+            #mp_drawing.draw_detection(img, detection)
+            #print(detection)
+            # detected_face is the cropped image that is then passed forward to the Regognizer
+            '''
+            DETECTION - 
+            Collection of detected faces, where each face is represented as a detection proto message that contains 
+            a bounding box and 6 key points (right eye, left eye, nose tip, mouth center, right ear tragion, and left
+            ear tragion). The bounding box is composed of xmin and width (both normalized to [0.0, 1.0] by the
+            image width) and ymin and height (both normalized to [0.0, 1.0] by the image height). Each key point
+            is composed of x and y, which are normalized to [0.0, 1.0] by the image width and height
+            respectively.
+            '''
+            # Bounding Box
+            x = re.findall('xmin: (..*)',str(detection))
+            y = re.findall('ymin: (..*)',str(detection))
+            h = re.findall('height: (..*)',str(detection))
+            w = re.findall('width: (..*)',str(detection))
+            # Eye Locations
+            reye_x = re.findall('x: (..*)',str(detection))[0]
+            leye_x = re.findall('x: (..*)',str(detection))[1]
+            reye_y = re.findall('y: (..*)', str(detection))[0]
+            leye_y = re.findall('y: (..*)', str(detection))[1]
+            # Detections are normalized by the mediapipe API, thus they need to be multiplied
+            # Extra tweaking done to improve accuracy
+            x = (float(x[0]) * original_size[1])
+            y = (float(y[0]) * original_size[0]-15)
+            h = (float(h[0]) * original_size[0]+10)
+            w = (float(w[0]) * original_size[1]+10)
+            reye_x = (float(reye_x) * original_size[1])
+            leye_x = (float(leye_x) * original_size[1])
+            reye_y = (float(reye_y) * original_size[0])
+            leye_y = (float(leye_y) * original_size[0])
+            if float(x) and float(y) > 0:
+                detected_face = img[int(y):int(y + h), int(x):int(x + w)]
+                img_region = [int(x), int(y), int(w), int(h)]
+                if align:
+                    left_eye=(leye_x,leye_y)
+                    right_eye=(reye_x,reye_y)
+                    #print(left_eye)
+                    #print(right_eye)
+                    detected_face = FaceDetector.alignment_procedure(detected_face, left_eye, right_eye)
+                resp.append((detected_face,img_region))
+            else:
+                continue
+
+            #print("Yahoo")
+    return resp
+
+
+#face_detector = FaceDetector.build_model('mediapipe')

requirements.txt

@@ -12,3 +12,4 @@ mtcnn>=0.1.0
 lightgbm>=2.3.1
 dlib>=19.20.0
 retina-face>=0.0.1
+mediapipe>=0.8.7.3