affichage des signaux faciaux en overlay

visualisation_live.py

@@ -3,7 +3,6 @@ import mediapipe as mp
 import numpy as np
 import sys
 
-# Indices landmarks
 NOSE_TIP = 1
 CHIN = 152
 FOREHEAD = 10
@@ -11,6 +10,12 @@ LEFT_EYE_OUTER = 263
 RIGHT_EYE_OUTER = 33
 LEFT_MOUTH = 61
 RIGHT_MOUTH = 291
+TOP_LIP = 13
+BOTTOM_LIP = 14
+LEFT_EYE_TOP = 159
+LEFT_EYE_BOTTOM = 145
+RIGHT_EYE_TOP = 386
+RIGHT_EYE_BOTTOM = 374
 
 FACE_3D_MODEL = np.array([
     [0.0, 0.0, 0.0],
@@ -50,7 +55,7 @@ def dessiner_axe_tete(frame, landmarks, frame_w, frame_h):
         FACE_3D_MODEL, image_points, camera_matrix, np.zeros((4, 1))
     )
     if not success:
-        return
+        return None, None, None
 
     axis_length = 80
     axes_3d = np.float32([
@@ -65,6 +70,38 @@ def dessiner_axe_tete(frame, landmarks, frame_w, frame_h):
     cv2.line(frame, nose, tuple(axes_2d[1].ravel().astype(int)), (0, 255, 0), 2)
     cv2.line(frame, nose, tuple(axes_2d[2].ravel().astype(int)), (255, 0, 0), 2)
 
+    rotation_mat, _ = cv2.Rodrigues(rotation_vec)
+    angles, _, _, _, _, _ = cv2.RQDecomp3x3(rotation_mat)
+    return angles[0], angles[1], angles[2]
+
+
+def dessiner_signaux(frame, lm, pitch, yaw, roll):
+    face_height = abs(lm[CHIN].y - lm[FOREHEAD].y)
+    face_width = abs(lm[LEFT_EYE_OUTER].x - lm[RIGHT_EYE_OUTER].x)
+
+    mouth_opening = abs(lm[TOP_LIP].y - lm[BOTTOM_LIP].y) / face_height if face_height > 0 else 0
+    left_eye = abs(lm[LEFT_EYE_TOP].y - lm[LEFT_EYE_BOTTOM].y) / face_height if face_height > 0 else 0
+    right_eye = abs(lm[RIGHT_EYE_TOP].y - lm[RIGHT_EYE_BOTTOM].y) / face_height if face_height > 0 else 0
+    smile = abs(lm[LEFT_MOUTH].x - lm[RIGHT_MOUTH].x) / face_width if face_width > 0 else 0
+
+    lignes = [
+        f"Pitch : {pitch:.1f}",
+        f"Yaw   : {yaw:.1f}",
+        f"Roll  : {roll:.1f}",
+        f"Bouche: {mouth_opening:.3f}",
+        f"Oeil G: {left_eye:.3f}",
+        f"Oeil D: {right_eye:.3f}",
+        f"Sourire: {smile:.3f}",
+    ]
+
+    overlay = frame.copy()
+    cv2.rectangle(overlay, (8, 8), (200, 18 + len(lignes) * 20), (0, 0, 0), -1)
+    cv2.addWeighted(overlay, 0.5, frame, 0.5, 0, frame)
+
+    for i, texte in enumerate(lignes):
+        cv2.putText(frame, texte, (12, 24 + i * 20),
+                    cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 180), 1)
+
 
 def visualiser_live(chemin_video):
     cap = cv2.VideoCapture(chemin_video)
@@ -96,7 +133,10 @@ def visualiser_live(chemin_video):
                 lm = face_landmarks.landmark
 
                 dessiner_landmarks(frame, face_landmarks, frame_w, frame_h)
-                dessiner_axe_tete(frame, lm, frame_w, frame_h)
+                pitch, yaw, roll = dessiner_axe_tete(frame, lm, frame_w, frame_h)
+
+                if pitch is not None:
+                    dessiner_signaux(frame, lm, pitch, yaw, roll)
 
             cv2.imshow("Visualisation faciale - Q pour quitter", frame)
             if cv2.waitKey(wait_ms) & 0xFF == ord('q'):