Merge pull request akshitagupta15june#799 from iamakkkhil/iamakkkhil-…

…Wolverine

Wolverine Filter added

Snapchat_Filters/Wolverine_Filter/README.md

@@ -0,0 +1,20 @@
+# <b> Wolverine Superhero Snapchat Filter </b>
+
+## Setting up and running the script
+
+- Install the requirements of the script as follows
+```
+    $ pip install -r requirements.txt
+```
+
+- Run the script as follows
+```
+    $ python Wolverine_filter.py
+```
+
+## Output screenshot
+
+![Image](https://iili.io/B6bAJV.md.png)
+
+## Author
+[Akhil Bhalerao](https://github.com/iamakkkhil)

Snapchat_Filters/Wolverine_Filter/Wolverine_Filter.py

@@ -0,0 +1,72 @@
+import dlib
+import cv2
+
+
+def mask(frame, landmarks):
+    imgWolvie = cv2.imread("wolvie.png", -1)
+    orig_mask = imgWolvie[:, :, 3]
+    orig_mask_inv = cv2.bitwise_not(orig_mask)
+    imgWolvie = imgWolvie[:, :, 0:3]
+    origWolvieHeight, origWolvieWidth = imgWolvie.shape[:2]
+
+    WolvieWidth = abs(3 * (landmarks.part(37).x - landmarks.part(44).x))
+    WolvieHeight = int(WolvieWidth * origWolvieHeight / origWolvieWidth)
+    Wolvie = cv2.resize(imgWolvie, (WolvieWidth, WolvieHeight), interpolation=cv2.INTER_AREA)
+    mask = cv2.resize(orig_mask, (WolvieWidth, WolvieHeight), interpolation=cv2.INTER_AREA)
+    mask_inv = cv2.resize(orig_mask_inv, (WolvieWidth, WolvieHeight), interpolation=cv2.INTER_AREA)
+    y1 = int(landmarks.part(27).y - (WolvieHeight / 2)) - 50
+    y2 = int(y1 + WolvieHeight)
+    x1 = int(landmarks.part(27).x - (WolvieWidth / 2))
+    x2 = int(x1 + WolvieWidth)
+    roi = frame[y1:y2, x1:x2]
+    roi_bg = cv2.bitwise_and(roi, roi, mask=mask_inv)
+    roi_fg = cv2.bitwise_and(Wolvie, Wolvie, mask=mask)
+    frame[y1:y2, x1:x2] = cv2.add(roi_bg, roi_fg)
+
+    return frame
+
+
+def filter():
+    """
+    This function consists main logic of the program in which
+    1. detect faces
+    2. from 68 landmark points we detect eyes
+    3. Applies Filter
+    :return: None
+    """
+
+    # detector for detecting the face in the image
+    detector = dlib.get_frontal_face_detector()
+    # predictor of locating 68 landmark points from the face by using a pretrained model
+    predictor = dlib.shape_predictor('shape_predictor_68_face_landmarks.dat')
+
+    cap = cv2.VideoCapture(0)
+    while True:
+        ret, frame = cap.read()
+        if ret:
+            frameGray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
+            # detecting faces in the frame
+            faces = detector(frameGray)
+
+            # if faces are present then locating the landmark points
+            for face in faces:
+                landmarks = predictor(frameGray, face)
+
+                frame = mask(frame, landmarks)
+
+            # for showing frames on the window named Detector
+            cv2.imshow('Detector', frame)
+
+            # for quiting the program press 'ESC'
+            if cv2.waitKey(1) & 0xFF == 27:
+                break
+        else:
+            break
+
+    # releasing all the frames we captured and destroying the windows
+    cap.release()
+    cv2.destroyAllWindows()
+
+
+if __name__ == '__main__':
+    filter()

Snapchat_Filters/Wolverine_Filter/requirements.txt

@@ -0,0 +1,4 @@
+dlib==19.21.1
+numpy==1.20.1
+opencv-python==4.5.1.48
+scipy==1.6.1