-
Notifications
You must be signed in to change notification settings - Fork 30
/
GaitDemoV1.py
240 lines (220 loc) · 10.3 KB
/
GaitDemoV1.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
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
"""
Gait demo for Tutorial V1.0
Author: Chunfeng Song
E-mail: [email protected]
This gait recognition demo needs clean background, e.g. a white wall. As this demo only takes GEI as its feature, so it can only recognize the same view. For example, people walk with the same angle for both registration and recognition.
"""
# Make sure you have install the following Python packages, we recommend you to install Anaconda and OpencCV 2.4. This code can work in both Windows and Linux.
import sys
import cPickle
from PyQt4 import QtCore, QtGui, uic
import numpy as n
from PIL import Image
from PIL.ImageQt import ImageQt
import cv2
import os
#Loading the UI window
qtCreatorFile = "GaitUI.ui"
Ui_MainWindow, QtBaseClass = uic.loadUiType(qtCreatorFile)
def pickle(filename, data, compress=False):
fo = open(filename, "wb")
cPickle.dump(data, fo, protocol=cPickle.HIGHEST_PROTOCOL)
fo.close()
def unpickle(filename):
fo = open(filename, 'rb')
dict = cPickle.load(fo)
fo.close()
return dict
class GaitDemo(QtGui.QMainWindow, Ui_MainWindow):
def __init__(self):
'''
Set initial parameters here.
Note that the demo window size is 1366*768, you can edit this via Qtcreator.
In this demo, we take 20 frames of profiles to generate a GEI. You can edit this number by your self.
'''
QtGui.QMainWindow.__init__(self)
Ui_MainWindow.__init__(self)
self.showFullScreen()
self.setupUi(self)
self.capture = cv2.VideoCapture(0)# Edit this default num to 1 or 2, if you have multiple cameras.
self.currentFrame=n.array([])
self.firstFrame=None
self.register_state = False
self.recognition_state = False
self.save_on = False
self.gei_fix_num = 20
#Ok, this is our logos, welcome to contact and visit us.
self.logo1 = QtGui.QLabel(self.centralwidget)
self.logo1.setPixmap(QtGui.QPixmap(QtGui.QPixmap('logo1.png').scaled(250,70)))
self.logo1.setGeometry(5, 698, 250, 70)
self.logo2 = QtGui.QLabel(self.centralwidget)
self.logo2.setPixmap(QtGui.QPixmap(QtGui.QPixmap('logo2.png').scaled(170,70)))
self.logo2.setGeometry(275, 698, 170, 70)
self.logo3 = QtGui.QLabel(self.centralwidget)
self.logo3.setPixmap(QtGui.QPixmap(QtGui.QPixmap('logo3.png').scaled(243,70)))
self.logo3.setGeometry(480, 698, 243, 70)
#Set two window for raw video and segmentation.
self.video_lable=QtGui.QLabel(self.centralwidget)
self.seg_label = QtGui.QLabel(self.centralwidget)
self._timer = QtCore.QTimer(self)
self.video_lable.setGeometry(50,100, 512, 384)
self.seg_label.setGeometry(612,100, 512, 384)
self.load_dataset()
self._timer.timeout.connect(self.play)
#Waiting for you to push the button.
self.register_2.clicked.connect(self.register_show)
self.recognize.clicked.connect(self.recognition_show)
self.updater.clicked.connect(self.update_bk)
self.save_gei.clicked.connect(self.save_gei_f)
self._timer.start(27)
self.update()
def save_gei_f(self):
'''
Waiting the save button.
'''
self.save_on = True
self.state_print.setText('Saving!')
def register_show(self):
'''
To record the GEI into gait database.
'''
self.register_state = True
self.recognition_state = False
self.state_print.setText('Register!')
self.gei_current = n.zeros((128,88), n.single)
self.numInGEI = 0
def load_dataset(self):
'''
Load gait database if existing.
'''
self.data_path = './GaitData'
if os.path.exists(self.data_path):
dic = unpickle(self.data_path)
self.num = dic['num']
self.gei = dic['gei']
self.name = dic['name']
else:
self.num = 0
self.gei = n.zeros([100,128,88],n.uint8)
self.name = []
dic = {'num':self.num, 'gei':self.gei, 'name':self.name}
pickle(self.data_path, dic, compress=False)
self.id_num.setText('%d' %self.num)
self.state_print.setText('Running!')
def recognition_show(self):
'''
Working now and just recognizing the one in front of this camera.
'''
self.recognition_state = True
self.register_state = False
self.gei_current = n.zeros((128,88), n.single)
self.numInGEI = 0
self.state_print.setText('Recognition!')
def update_bk(self):
'''
If you moved the camera.
'''
self.firstFrame = self.FrameForUpdate
def play(self):
'''
Main program.
'''
ret, frame=self.capture.read() #Read video from a camera.
if(ret==True):
frame = cv2.resize(frame,(512,384))
#Apply background subtraction method.
gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
gray = cv2.GaussianBlur(gray, (3,3), 0)
if self.firstFrame is None:
self.firstFrame = gray # Set this frame as the background.
frameDelta = cv2.absdiff(self.firstFrame, gray)
thresh = cv2.threshold(frameDelta, 50, 255, cv2.THRESH_BINARY)[1]
self.FrameForUpdate = gray
thresh = cv2.dilate(thresh, None, iterations=2)
(cnts, _) = cv2.findContours(thresh.copy(), cv2.RETR_EXTERNAL,
cv2.CHAIN_APPROX_SIMPLE)
thresh = n.array(thresh)
max_rec=0
#Find the max box.
for c in cnts:
if cv2.contourArea(c) < 500:
continue
(x, y, w, h) = cv2.boundingRect(c)
if w>25 and h>50:
if max_rec<w*h:
max_rec = w*h
(x_max, y_max, w_max, h_max) = cv2.boundingRect(c)
#If exist max box.
if max_rec>0:
cv2.rectangle(frame, (x_max, y_max), (x_max + w_max, y_max + h_max), (0, 255, 0), 2)
if x_max>20: #To ignore some regions which contain parts of human body.
if self.register_state or self.recognition_state:
nim = n.zeros([thresh.shape[0]+10,thresh.shape[1]+10],n.single) # Enlarge the box for better result.
nim[y_max+5:(y_max + h_max+5),x_max+5:(x_max + w_max+5)] = thresh[y_max:(y_max + h_max),x_max:(x_max + w_max)]
offsetX = 20
# Get coordinate position.
ty, tx = (nim >100).nonzero()
sy, ey = ty.min(), ty.max()+1
sx, ex = tx.min(), tx.max()+1
h = ey - sy
w = ex - sx
if h>w:# Normal human should be like this, the height shoud be greater than wideth.
# Calculate the frame for GEI
cx = int(tx.mean())
cenX = h/2
start_w = (h-w)/2
if max(cx-sx,ex-cx)<cenX:
start_w = cenX - (cx-sx)
tim = n.zeros((h,h), n.single)
tim[:,start_w:start_w+w] = nim[sy:ey,sx:ex]
rim = Image.fromarray(n.uint8(tim)).resize((128,128), Image.ANTIALIAS)
tim = n.array(rim)[:,offsetX:offsetX+88]
if self.numInGEI<self.gei_fix_num:
self.gei_current += tim # Add up until reaching the fix number.
self.numInGEI += 1
if self.numInGEI>self.gei_fix_num:
if self.save_on:
#Save the GEI.
self.gei[self.num,:,:] = self.gei_current/self.gei_fix_num
Image.fromarray(n.uint8(self.gei_current/self.gei_fix_num)).save('./gei/gei%02d%s.jpg'%(self.num,self.id_name.toPlainText()))
self.name.append(self.id_name.toPlainText())
self.num +=1
self.id_num.setText('%d' %self.num)
dic = {'num':self.num, 'gei':self.gei, 'name':self.name}
pickle(self.data_path, dic, compress=False)
self.save_on = False
self.state_print.setText('Saved!')
elif self.recognition_state:
#Recognition.
self.gei_query = self.gei_current/(self.gei_fix_num)
score = n.zeros(self.num)
self.gei_to_com = n.zeros([128,88],n.single)
for q in xrange(self.num):
self.gei_to_com = self.gei[q,:,:]
score[q]=n.exp(-(((self.gei_query[:]-self.gei_to_com[:])/(128*88))**2).sum())#Compare with gait database.
q_id = score.argmax()
if True:
id_rec = '%s' %self.name[q_id]
cv2.putText(frame,id_rec,(x_max+20,y_max+20),fontFace=cv2.FONT_HERSHEY_SIMPLEX, fontScale=1.0, thickness=2,color=(0,0,255))
else:
self.gei_current = n.zeros((128,88), n.single)
self.numInGEI = 0
#Show results.
self.currentFrame=cv2.cvtColor(frame,cv2.COLOR_BGR2RGB)
self.currentSeg=Image.fromarray(thresh).convert('RGB')
self.currentSeg = ImageQt(self.currentSeg)
height,width=self.currentFrame.shape[:2]
img=QtGui.QImage(self.currentFrame,
width,
height,
QtGui.QImage.Format_RGB888)
img=QtGui.QPixmap.fromImage(img)
self.video_lable.setPixmap(img)
seg=QtGui.QImage(self.currentSeg)
seg=QtGui.QPixmap(seg)
self.seg_label.setPixmap(seg)
if __name__ == "__main__":
app = QtGui.QApplication(sys.argv)
window = GaitDemo()
window.show()
sys.exit(app.exec_())