more cleanups

faceit_live.py

@@ -96,7 +96,6 @@ def main():
     print("webcam dimensions = {} x {}".format(width,height))
 
     # load models
-    previous = None
     net = load_face_model()
     generator, kp_detector = demo.load_checkpoints(config_path=f'{first_order_path}config/vox-adv-256.yaml', checkpoint_path=f'{model_path}/vox-adv-cpk.pth.tar')
 
@@ -108,9 +107,7 @@ def main():
 
     cv2.namedWindow('DeepFake', cv2.WINDOW_GUI_NORMAL) # face transformation
     cv2.moveWindow('DeepFake', int(screen_width//2)+150, 100)
-    img_shape = source_image.shape
-    cv2.resizeWindow('DeepFake', int(img_shape[1] // img_shape[0] * 256), 256)
-
+    cv2.resizeWindow('DeepFake', 256, 256)
 
     cv2.namedWindow('Stream', cv2.WINDOW_GUI_NORMAL) # rendered to fake webcam
     cv2.moveWindow('Stream', int(screen_width//2)-int(webcam_width//2), 400)
@@ -120,6 +117,8 @@ def main():
     print("Press C to center Webcam, Press B/N for previous/next image in media directory, T to alter between relative and absolute transformation, Q to quit")
     x1,y1,x2,y2 = [0,0,0,0]
     relative = True
+    previous = None
+
     while True:
         ret, frame = video_capture.read()
         frame = cv2.resize(frame, (640, 480))
@@ -134,17 +133,13 @@ def main():
             #cv2.imshow('Previous',previous)
 
 
-        curr_face = cut_face_window(x1,y1,x2,y2,frame)
-        cv2.imshow('Previous',previous)
-        cv2.imshow('Curr Face',curr_face)
-        cv2.imshow('Source Image',source_image)
-
-        print('previous=',previous.shape)
-        print('curr_face=',curr_face.shape)
-        print('source=',source_image.shape)
+        curr_face = cut_face_window(x1,y1,x2,y2,frame.copy())
+        # cv2.imshow('Previous',previous)
+        # cv2.imshow('Curr Face',curr_face)
+        # cv2.imshow('Source Image',source_image)
 
         deep_fake = process_image(source_image,previous,curr_face,net, generator, kp_detector, relative)
-        print("deep_fake",deep_fake.shape)
+        #print("deep_fake",deep_fake.shape)
 
         deep_fake = cv2.cvtColor(deep_fake, cv2.COLOR_RGB2BGR) 
 
@@ -154,26 +149,23 @@ def main():
         x_border = int((640-(img_shape[1] // img_shape[0] * 480))//2)
         #y_border = int((480-(img_shape[0] // img_shape[1] * 640))//2)
 
-
-
-
         stream_v = cv2.copyMakeBorder(rgb, 0, 0, x_border if x_border >=0 else 0, x_border if x_border >=0 else 0, cv2.BORDER_CONSTANT)
         
-        cv2.imshow('Webcam', frame)
+        #cv2.imshow('Webcam', frame)
         cv2.imshow('Face', curr_face)
         cv2.imshow('DeepFake', deep_fake)
         #cv2.imshow('Previous', previous)
         #cv2.imshow('RGB', rgb)
         #cv2.imshow('Source Image', source_image)
-        cv2.imshow('Stream',stream_v)
-        
         #time.sleep(1/30.0)
-        stream_v = cv2.flip(stream_v,1)
-        stream_v = cv2.cvtColor(stream_v, cv2.COLOR_BGR2RGB)
-        stream_v = (stream_v*255).astype(np.uint8)
+
+        cv2.imshow('Stream',stream_v)
 
         # stream to fakewebcam
         if system=="linux":
+            stream_v = cv2.flip(stream_v,1)
+            stream_v = cv2.cvtColor(stream_v, cv2.COLOR_BGR2RGB)
+            stream_v = (stream_v*255).astype(np.uint8)
             #print("output to fakecam")
             camera.schedule_frame(stream_v)
 
@@ -219,12 +211,13 @@ def load_face_model():
     return net
 
 def cut_face_window(x1,y1,x2,y2,frame):
+    frame = frame.copy()
     frame = frame[y1:y2,x1:x2]
-    face = resize(frame, (256, 256))# [..., :3]
+    face = resize(frame, (256, 256))[..., :3]
     return face
 
 # find the face in webcam stream and center a 256x256 window
-def find_face_cut(net,face,previous=False):
+def find_face_cut(net,face):
     blob = cv2.dnn.blobFromImage(face, 1.0, (300, 300), [104, 117, 123], False, False)
     frameWidth = 640
     frameHeight = 480
@@ -233,31 +226,33 @@ def find_face_cut(net,face,previous=False):
     bboxes = []
     face_found = False
     for i in range(detections.shape[2]):
+        #print(i)
         confidence = detections[0, 0, i, 2]
-        if confidence > 0.8:
-            face_found = True
-            x1 = int(detections[0, 0, i, 3] * frameWidth)
-            y1 = int(detections[0, 0, i, 4] * frameHeight)
-            x2 = int(detections[0, 0, i, 5] * frameWidth)
-            y2 = int(detections[0, 0, i, 6] * frameHeight)
+        if confidence > 0.9:
+            x1 = (int(detections[0, 0, i, 3] * frameWidth)//2)*2
+            y1 = (int(detections[0, 0, i, 4] * frameHeight)//2)*2
+            x2 = (int(detections[0, 0, i, 5] * frameWidth)//2)*2
+            y2 = (int(detections[0, 0, i, 6] * frameHeight)//2)*2
 
             face_margin_w = int(256 - (abs(x1-x2)))
             face_margin_h = int(256 - (abs(y1-y2)))
 
-            cut_x1 = (x1 - int(face_margin_w//2))
-            if cut_x1<0: cut_x1=0
+            cut_x1 = x1 - int(face_margin_w//2)
             cut_y1 = y1 - int(2*face_margin_h//3)
-            if cut_y1<0: cut_y1=0
+
             cut_x2 = x2 + int(face_margin_w//2)
-            cut_y2 = y2 + int(face_margin_h//3)
+            cut_y2 = y2 + face_margin_h - int(2*face_margin_h//3)
+
+            face_found = True
+            break
 
     if not face_found:
         print("No face detected in video")
         # let's just use the middle section of the image
         cut_x1,cut_y1,cut_x2,cut_y2 = 112,192,368,448
     else:
-        print(f'Found face at: ({x1,y1}) ({x2},{y2} width:{abs(x2-x1)} height: {abs(y2-y1)})')
-        print(f'Cutting at: ({cut_x1,cut_y1}) ({cut_x2},{cut_y2} width:{abs(cut_x2-cut_x1)} height: {abs(cut_y2-cut_y1)})')
+        print(f'Found face at: ({x1,y1}) ({x2},{y2} width:{(x2-x1)} height: {(y2-y1)})')
+        print(f'Cutting at: ({cut_x1,cut_y1}) ({cut_x2},{cut_y2} width:{(cut_x2-cut_x1)} height: {(cut_y2-cut_y1)})')
 
 
     return cut_x1,cut_y1,cut_x2,cut_y2