File size: 1,455 Bytes
a94c5c8
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
88e55d3
 
a94c5c8
 
 
88e55d3
 
a94c5c8
 
 
 
 
88e55d3
a94c5c8
 
 
 
 
 
88e55d3
a94c5c8
 
88e55d3
a94c5c8
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
import os
import cv2
import numpy as np
from insightface.app import FaceAnalysis

def check_liveness_on_folder(folder_path, frame_interval=1, yaw_thresh=5, pitch_thresh=5):
    model = FaceAnalysis(name='buffalo_l', providers=['CPUExecutionProvider'])
    model.prepare(ctx_id=0)

    poses = []
    image_files = sorted([
        f for f in os.listdir(folder_path)
        if f.lower().endswith((".jpg", ".jpeg", ".png"))
    ])

    for idx, filename in enumerate(image_files):
        if idx % frame_interval != 0:
            continue

        img_path = os.path.join(folder_path, filename)
        img = cv2.imread(img_path)
        if img is None:
            print(f"Image illisible : {filename}")
            return False

        faces = model.get(img)
        if not faces:
            print(f"Aucun visage détecté sur {filename}")
            return False

        pose = faces[0].pose  # [yaw, pitch, roll]
        poses.append(pose)

    if len(poses) < 3:
        print("Pas assez de données pour vérifier la vivacité.")
        return False

    poses = np.array(poses)
    yaw_std = np.std(poses[:, 0])
    pitch_std = np.std(poses[:, 1])

    print(f"Écarts-types — Yaw: {yaw_std:.2f}, Pitch: {pitch_std:.2f}")

    if yaw_std < yaw_thresh and pitch_std < pitch_thresh:
        print("Visage figé détecté — FAKE")
        return False
    else:
        print("🟢 Mouvement naturel détecté — REAL")
        return True