Counting unique faces in video

computer-vision/unique_face_count_in_video.py

@@ -0,0 +1,44 @@
+import cv2
+face_cascade = cv2.CascadeClassifier('/home/spartan/Desktop/opencv-master/data/haarcascades/haarcascade_frontalface_default.xml')
+
+from skimage.measure import compare_ssim,compare_mse
+
+#used mse to compare two images but there are also many other options available at http://scikit-image.org/docs/dev/api/skimage.measure.html
+import math
+
+videoFile = "faces.mp4"
+vidcap = cv2.VideoCapture(videoFile)
+success,image = vidcap.read()
+previousframe=None
+
+
+
+seconds = 1
+fps = vidcap.get(cv2.CAP_PROP_FPS) 
+print(fps)
+multiplier = round(fps * seconds)
+print(multiplier)
+faces = face_cascade.detectMultiScale(image, 1.3, 5)
+count=len(faces)
+while success:
+    if(previousframe is None):
+        previousframe=image
+    frameId = int(round(vidcap.get(1))) 
+    success,image = vidcap.read()
+
+    if frameId % multiplier == 0:
+        score=compare_mse(image,previousframe)
+        if(score>4000):
+            faces=face_cascade.detectMultiScale(image, 1.3, 5)
+            print (len(faces))
+            print(frameId)
+            count=count+len(faces)
+        previousframe=image    
+        cv2.putText(image,('%d unique faces found till now' % count),(250,300), cv2.FONT_HERSHEY_SIMPLEX,1, (200,255,155), 2)
+
+        cv2.imshow('ip',image)
+        cv2.waitKey(30)
+        cv2.imwrite("/home/spartan/nn/frame%d.jpg" % frameId, image)
+
+vidcap.release()
+print ("Complete")