Merge branch 'master' into regions

README.md

@@ -301,7 +301,9 @@ You can also support this work on [Patreon](https://www.patreon.com/serengil?rep
 
 ## Citation
 
-Please cite deepface in your publications if it helps your research for facial recognition purposes. Here are its BibTex entries:
+Please cite deepface in your publications if it helps your research. Here are its BibTex entries:
+
+If you use deepface for facial recogntion purposes, please cite the this publication.
 
 ```BibTeX
 @inproceedings{serengil2020lightface,
@@ -316,7 +318,7 @@ Please cite deepface in your publications if it helps your research for facial r
 }
 ```
 
-If you use deepface for facial attribute analysis purposes such as age, gender, emotion or ethnicity, please cite the this publication.
+If you use deepface for facial attribute analysis purposes such as age, gender, emotion or ethnicity prediction, please cite the this publication.
 
 ```BibTeX
 @inproceedings{serengil2021lightface,

deepface/DeepFace.py

@@ -395,7 +395,7 @@ def analyze(img_path, actions = ('emotion', 'age', 'gender', 'race') , models =
 				emotion_labels = ['angry', 'disgust', 'fear', 'happy', 'sad', 'surprise', 'neutral']
 				img, region = functions.preprocess_face(img = img_path, target_size = (48, 48), grayscale = True, enforce_detection = enforce_detection, detector_backend = detector_backend, return_region = True)
 
-				emotion_predictions = models['emotion'].predict(img)[0,:]
+				emotion_predictions = models['emotion'].predict(img, verbose=0)[0,:]
 
 				sum_of_predictions = emotion_predictions.sum()
 
@@ -412,7 +412,7 @@ def analyze(img_path, actions = ('emotion', 'age', 'gender', 'race') , models =
 				if img_224 is None:
 					img_224, region = functions.preprocess_face(img = img_path, target_size = (224, 224), grayscale = False, enforce_detection = enforce_detection, detector_backend = detector_backend, return_region = True)
 
-				age_predictions = models['age'].predict(img_224)[0,:]
+				age_predictions = models['age'].predict(img_224, verbose=0)[0,:]
 				apparent_age = Age.findApparentAge(age_predictions)
 
 				resp_obj["age"] = int(apparent_age) #int cast is for the exception - object of type 'float32' is not JSON serializable
@@ -422,7 +422,7 @@ def analyze(img_path, actions = ('emotion', 'age', 'gender', 'race') , models =
 				if img_224 is None:
 					img_224, region = functions.preprocess_face(img = img_path, target_size = (224, 224), grayscale = False, enforce_detection = enforce_detection, detector_backend = detector_backend, return_region = True)
 
-				gender_predictions = models['gender'].predict(img_224)[0,:]
+				gender_predictions = models['gender'].predict(img_224, verbose=0)[0,:]
 
 				gender_labels = ["Woman", "Man"]
 				resp_obj["gender"] = {}
@@ -436,7 +436,7 @@ def analyze(img_path, actions = ('emotion', 'age', 'gender', 'race') , models =
 			elif action == 'race':
 				if img_224 is None:
 					img_224, region = functions.preprocess_face(img = img_path, target_size = (224, 224), grayscale = False, enforce_detection = enforce_detection, detector_backend = detector_backend, return_region = True) #just emotion model expects grayscale images
-				race_predictions = models['race'].predict(img_224)[0,:]
+				race_predictions = models['race'].predict(img_224, verbose=0)[0,:]
 				race_labels = ['asian', 'indian', 'black', 'white', 'middle eastern', 'latino hispanic']
 
 				sum_of_predictions = race_predictions.sum()
@@ -765,7 +765,12 @@ def represent(img_path, model_name = 'VGG-Face', model = None, enforce_detection
 	#---------------------------------
 
 	#represent
-	embedding = model.predict(img)[0].tolist()
+	if "keras" in str(type(model)):
+		#new tf versions show progress bar and it is annoying
+		embedding = model.predict(img, verbose=0)[0].tolist()
+	else:
+		#SFace is not a keras model and it has no verbose argument
+		embedding = model.predict(img)[0].tolist()
 
 	return embedding
 

deepface/commons/functions.py

@@ -110,7 +110,7 @@ def detect_face(img, detector_backend = 'opencv', grayscale = False, enforce_det
 	face_detector = FaceDetector.build_model(detector_backend)
 
 	try:
-		detected_face, img_region = FaceDetector.detect_face(face_detector, detector_backend, img, align)
+		detected_face, img_region, _ = FaceDetector.detect_face(face_detector, detector_backend, img, align)
 	except: #if detected face shape is (0, 0) and alignment cannot be performed, this block will be run
 		detected_face = None
 

deepface/commons/realtime.py

@@ -151,7 +151,7 @@ def analysis(db_path, model_name = 'VGG-Face', detector_backend = 'opencv', dist
 
 		detected_faces = []
 		face_index = 0
-		for face, (x, y, w, h) in faces:
+		for face, (x, y, w, h), _ in faces:
 			if w > 130: #discard small detected faces
 
 				face_detected = True

deepface/detectors/DlibWrapper.py

@@ -45,10 +45,13 @@ def detect_face(detector, img, align = True):
 	sp = detector["sp"]
 
 	detected_face = None
+
 	img_region = [0, 0, img.shape[1], img.shape[0]]
 
 	face_detector = detector["face_detector"]
-	detections = face_detector(img, 1)
+
+	#note that, by design, dlib's fhog face detector scores are >0 but not capped at 1
+	detections, scores, _ = face_detector.run(img, 1)
 
 	if len(detections) > 0:
 
@@ -60,12 +63,13 @@ def detect_face(detector, img, align = True):
 			detected_face = img[max(0, top): min(bottom, img.shape[0]), max(0, left): min(right, img.shape[1])]
 			
 			img_region = [left, top, right - left, bottom - top]
+			confidence = scores[idx]
 
 			if align:
 				img_shape = sp(img, detections[idx])
 				detected_face = dlib.get_face_chip(img, img_shape, size = detected_face.shape[0])
 
-			resp.append((detected_face, img_region))
+			resp.append((detected_face, img_region, confidence))
 
 
 	return resp

deepface/detectors/FaceDetector.py

@@ -14,7 +14,7 @@ def build_model(detector_backend):
         'dlib': DlibWrapper.build_model,
         'mtcnn': MtcnnWrapper.build_model,
         'retinaface': RetinaFaceWrapper.build_model,
-	'mediapipe': MediapipeWrapper.build_model
+	    'mediapipe': MediapipeWrapper.build_model
     }
 
     if not "face_detector_obj" in globals():
@@ -37,12 +37,12 @@ def detect_face(face_detector, detector_backend, img, align = True):
     obj = detect_faces(face_detector, detector_backend, img, align)
 
     if len(obj) > 0:
-        face, region = obj[0] #discard multiple faces
+        face, region, confidence = obj[0] #discard multiple faces
     else: #len(obj) == 0
         face = None
         region = [0, 0, img.shape[1], img.shape[0]]
 
-    return face, region
+    return face, region, confidence
 
 def detect_faces(face_detector, detector_backend, img, align = True):
 
@@ -52,14 +52,14 @@ def detect_faces(face_detector, detector_backend, img, align = True):
         'dlib': DlibWrapper.detect_face,
         'mtcnn': MtcnnWrapper.detect_face,
         'retinaface': RetinaFaceWrapper.detect_face,
-	'mediapipe': MediapipeWrapper.detect_face
+	    'mediapipe': MediapipeWrapper.detect_face
     }
 
     detect_face = backends.get(detector_backend)
 
     if detect_face:
         obj = detect_face(face_detector, img, align)
-        #obj stores list of detected_face and region pair
+        #obj stores list of (detected_face, region, confidence)
 
         return obj
     else:

deepface/detectors/MediapipeWrapper.py

@@ -20,7 +20,7 @@ def detect_face(face_detector, img, align = True):
     if results.detections:
         for detection in results.detections:
 
-            confidence = detection.score
+            confidence, = detection.score
             
             bounding_box = detection.location_data.relative_bounding_box
             landmarks = detection.location_data.relative_keypoints
@@ -44,6 +44,6 @@ def detect_face(face_detector, img, align = True):
                 if align:
                     detected_face = FaceDetector.alignment_procedure(detected_face, left_eye, right_eye)
                     
-                resp.append((detected_face,img_region))
+                resp.append((detected_face, img_region, confidence))
                 
     return resp

deepface/detectors/MtcnnWrapper.py

@@ -22,6 +22,7 @@ def detect_face(face_detector, img, align = True):
 			x, y, w, h = detection["box"]
 			detected_face = img[int(y):int(y+h), int(x):int(x+w)]
 			img_region = [x, y, w, h]
+			confidence = detection["confidence"]
 
 			if align:
 				keypoints = detection["keypoints"]
@@ -29,6 +30,6 @@ def detect_face(face_detector, img, align = True):
 				right_eye = keypoints["right_eye"]
 				detected_face = FaceDetector.alignment_procedure(detected_face, left_eye, right_eye)
 
-			resp.append((detected_face, img_region))
+			resp.append((detected_face, img_region, confidence))
 
 	return resp

deepface/detectors/OpenCvWrapper.py

@@ -45,13 +45,15 @@ def detect_face(detector, img, align = True):
 	faces = []
 	try:
 		#faces = detector["face_detector"].detectMultiScale(img, 1.3, 5)
-		faces = detector["face_detector"].detectMultiScale(img, 1.1, 10)
+		
+		#note that, by design, opencv's haarcascade scores are >0 but not capped at 1
+		faces, _, scores = detector["face_detector"].detectMultiScale3(img, 1.1, 10, outputRejectLevels = True)
 	except:
 		pass
 
 	if len(faces) > 0:
 
-		for x,y,w,h in faces:
+		for (x,y,w,h), confidence in zip(faces, scores):
 			detected_face = img[int(y):int(y+h), int(x):int(x+w)]
 
 			if align:
@@ -59,7 +61,7 @@ def detect_face(detector, img, align = True):
 
 			img_region = [x, y, w, h]
 
-			resp.append((detected_face, img_region))
+			resp.append((detected_face, img_region, confidence))
 
 	return resp
 

deepface/detectors/RetinaFaceWrapper.py

@@ -44,6 +44,7 @@ def detect_face(face_detector, img, align = True):
             x = facial_area[0]
             w = facial_area[2] - x
             img_region = [x, y, w, h]
+            confidence = identity["score"]
 
             #detected_face = img[int(y):int(y+h), int(x):int(x+w)] #opencv
             detected_face = img[facial_area[1]: facial_area[3], facial_area[0]: facial_area[2]]
@@ -58,6 +59,6 @@ def detect_face(face_detector, img, align = True):
 
                 detected_face = postprocess.alignment_procedure(detected_face, right_eye, left_eye, nose)
 
-            resp.append((detected_face, img_region))
+            resp.append((detected_face, img_region, confidence))
 
     return resp

deepface/detectors/SsdWrapper.py

@@ -93,10 +93,11 @@ def detect_face(detector, img, align = True):
 
 			detected_face = base_img[int(top*aspect_ratio_y):int(bottom*aspect_ratio_y), int(left*aspect_ratio_x):int(right*aspect_ratio_x)]
 			img_region = [int(left*aspect_ratio_x), int(top*aspect_ratio_y), int(right*aspect_ratio_x) - int(left*aspect_ratio_x), int(bottom*aspect_ratio_y) - int(top*aspect_ratio_y)]
+			confidence = instance["confidence"]
 
 			if align:
 				detected_face = OpenCvWrapper.align_face(detector["eye_detector"], detected_face)
 
-			resp.append((detected_face, img_region))
+			resp.append((detected_face, img_region, confidence))
 
 	return resp