-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathpresidential.py
143 lines (100 loc) · 3.7 KB
/
presidential.py
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
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
#!/usr/bin/python2
import os, sys
import cv2
import numpy as np
from PIL import Image # Optional
fname = 'squirrel.png'
# Read the squirrel image
sq = cv2.imread(fname)
print sq.shape
cv2.imshow('squirrel', sq)
sq_gray = cv2.cvtColor(sq, cv2.COLOR_BGR2GRAY)
cv2.imshow('squirrel gray', sq_gray)
sq_gray_mask = sq_gray > 0
sq_mask = np.array([[[v, v, v] for v in row] for row in sq_gray_mask])
#cv2.waitKey()
# Optional - PIL conversions
pilsq = Image.open(fname)
# PIL -> cv2
sq2 = np.asarray(pilsq)
sq2 = cv2.cvtColor(sq2, cv2.COLOR_RGB2BGR)
# cv2 -> PIL
pilsq2 = cv2.cvtColor(sq, cv2.COLOR_BGR2RGB)
pilsq2 = Image.fromarray(pilsq2)
# Load face detection cascade
print 'loading face cascade...'
cas_dir = '/usr/share/opencv/haarcascades/'
cas_fname = 'haarcascade_frontalface_default.xml'
#cas_fname = 'haarcascade_frontalface_alt.xml'
cas = cv2.CascadeClassifier(os.path.join(cas_dir, cas_fname))
# Webcam viewing
print 'starting webcam...'
cam = cv2.VideoCapture(0)
ret, frame = cam.read()
assert ret
print 'webcam res:', frame.shape
while True:
ret, frame = cam.read()
# overlay the squirrel
np.copyto(frame[-sq.shape[0]:, 0:sq.shape[1], :], sq, where=sq_mask)
frame_gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
faces = cas.detectMultiScale(
frame_gray,
scaleFactor=1.1,
minNeighbors=5,
minSize=(30, 30),
flags = cv2.CASCADE_SCALE_IMAGE
# v2.4 -> cv2.cv.CV_HAAR_SCALE_IMAGE
)
#print faces
#print 'frame: %s - %s' % (frame.shape, frame.dtype)
for (x, y, w, h) in faces:
# First augment it a little, because the original rect isn't very good
#cv2.rectangle(frame, (x, y), (x+w, y+h), (0, 255, 0), 2)
x0 = x
y0 = y
w0 = w
h0 = h
# Extend the box a bit
#y -= int(h * 0.1)
h += int(h * 0.3)
x -= int(w * 0.1)
w += int(w * 0.2)
if y < 0 or y+h > frame.shape[0]:
continue
if x < 0 or x+w > frame.shape[1]:
continue
#cv2.rectangle(frame, (x, y), (x+w, y+h), (0, 255, 255), 2)
# Make more presidential colour
# Get skin values
frame_hsv = cv2.cvtColor(frame, cv2.COLOR_BGR2HSV)
face_hsv = frame_hsv[y:y+h, x:x+w, :]
hsv_min = np.array([0, 0, 40])
hsv_max = np.array([200, 120, 220])
face_inrange = cv2.inRange(face_hsv, hsv_min, hsv_max)
face_inrange = np.array([[[v, v, v] for v in row] for row in face_inrange], dtype=bool)
#print 'face_inrange: %s - %s' % (face_inrange.shape, face_inrange.dtype)
pres_skin = [0.2, 0.65, 1.7]
face_bgr = frame[y:y+h, x:x+w, :] * pres_skin
face_bgr = np.minimum(face_bgr, 255)
face_bgr = np.array(face_bgr, dtype='uint8')
#print 'face_bgr: %s - %s' % (face_bgr.shape, face_bgr.dtype)
np.copyto(frame[y:y+h, x:x+w, :], face_bgr, where=face_inrange)
# More presidential hair
hair = sq[30:,0:70]
hair = np.rot90(hair, 3)
hair_mask = sq_mask[30:,0:70]
hair_mask = np.rot90(hair_mask, 3)
hair_scale = w0 / float(hair.shape[1])
hair = cv2.resize(hair, None, fx=hair_scale, fy=hair_scale)
hair_mask = np.array(hair_mask, dtype='uint8')
hair_mask = cv2.resize(hair_mask, None, fx=hair_scale, fy=hair_scale)
hair_mask = np.array(hair_mask, dtype=bool)
hair = np.array(np.minimum((hair * 0.7) + 100, 255), dtype='uint8') # Optional
if y0 - hair.shape[0] < 0:
continue
np.copyto(frame[y0-hair.shape[0]:y0, x0:x0+hair.shape[1], :], hair, where=hair_mask)
cv2.imshow('cam', frame)
key = cv2.waitKey(1)
if key != -1:
break