Junction2018/glasses.py at master · Yorkio/Junction2018 · GitHub

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
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
import cv2
import os
import face_recognition as fr
from PIL import Image

def load_resources(path):
    files = os.listdir(path)
    templates = []
    for file_path in files:
        if '.png' in file_path:
            relative_path = os.path.join(path, file_path)
            templates.append(cv2.imread(relative_path, -1))
            print (templates[-1].shape)
    return templates

def add_glasses(img, glasses, rate, cnt):
    face_cascade = cv2.CascadeClassifier('/home/york_io/.local/lib/python3.6/site-packages/cv2/data/haarcascade_frontalface_default.xml')
    eye_cascade = cv2.CascadeClassifier('/home/york_io/.local/lib/python3.6/site-packages/cv2/data/haarcascade_eye.xml')
    face_locs = []
    face_locs = []
    hat_locs = []
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    face_locs = face_cascade.detectMultiScale(gray, 1.3, 5)
    for loc in face_locs:
        x, y, w, h = loc
        roi_gray = gray[y:y+h, x:x+w]
        eyes = eye_cascade.detectMultiScale(roi_gray)
        if (len(eyes) < 2):
            continue
        expected_y = h // 3
        best_y = 123456789
        for eye in eyes:
            if abs(y - expected_y) < abs(best_y - expected_y):
                best_y = eye[1]
        gw = int(w * 0.9)
        delta = (w - gw) // 2
        print(glasses.shape)
        gh = int(gw / glasses.shape[1] * glasses.shape[0])
        gx = x + int(delta * 1.2)
        gy = y + expected_y - int(0.15 * gh)
        if gx < 0 or gy < 0:
            continue
        glass_img = cv2.resize(glasses, (gw, gh))
        for yi in range(gy, gy + gh):
            for xi in range(gx, gx + gw):
                if glass_img[yi - gy, xi - gx, 3] != 0:
                    img[yi, xi] = glass_img[yi - gy, xi - gx, 0:3]
    return face_locs, hat_locs, img

def add_hat(img, hat_templ, rate, cnt):
    face_cascade = cv2.CascadeClassifier('/home/york_io/.local/lib/python3.6/site-packages/cv2/data/haarcascade_frontalface_default.xml')
    face_locs = []
    hat_locs = []
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    face_locs = face_cascade.detectMultiScale(gray, 1.3, 5)
    for loc in face_locs:
        width_scaled = int(loc[2] * 1.5)
        width_original = hat_templ.shape[1]
        scale_factor = width_scaled / width_original
        estimated_height = int(hat_templ.shape[0] * scale_factor / 1.5)
        delta = int(0.2 * estimated_height)
        x, y, w, h = loc
        hat_locs.append([y - estimated_height + delta, x + w, y + delta, x])
    for hat_loc in hat_locs:
        if hat_loc[0] < 0:
            continue
        t, r, b, l = hat_loc
        w = r - l
        h = b - t
        hat_img = cv2.resize(hat_templ, (w, h))
        for y in range(t, b):
            for x in range(l, r):
                if hat_img[y - t, x - l, 3] != 0:
                    img[y, x] = hat_img[y - t, x - l, 0:3]
    return face_locs, hat_locs, img


def show_webcam(templ, process, rate = 5):
    cam = cv2.VideoCapture(0)
    cnt = 0
    face_locs = []
    while True:
        ret_val, img = cam.read()
        face_locs, obj_locs, img = process(img, templ, rate, cnt)
        cv2.imshow('my webcam', img)
        if cv2.waitKey(1) == 27:
            break  # esc to quit
        cnt += 1
    cv2.destroyAllWindows()


def main():
    hat_templ = load_resources('hat_resources')
    glass_templ = load_resources('glass_resources')
    for t in glass_templ:
        show_webcam(t, add_glasses, rate=3)
    for t in hat_templ:
        show_webcam(t, add_hat, rate=3)


if __name__ == '__main__':
    main()