-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathfullfinalimg.py
274 lines (207 loc) · 6.6 KB
/
fullfinalimg.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
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
import numpy as np
import cv2
import matplotlib.pyplot as plt
from keras.models import load_model
from scipy.misc import imresize
def startpixels(wpixel, pic):
for ly in range(0,len(wpixel)-1):
if wpixel[ly]!=0:
break
for lx in range(0,120):
if pic[lx,ly]!=0:
break
for ry in range(len(wpixel)-1,-1,-1):
if wpixel[ry]!=0:
break
for rx in range(0,120):
if pic[rx,ry]!=0:
break
return ly,lx,ry,rx
#sliding window up to down
def slidingwindowdown( x, y, pic, colourpic):
recpx=[]
recpy=[]
while(True):
cv2.rectangle(colourpic,(y-10,x),(y+10,x+10),(0,255,0),1)
avgi=0
avgj=0
c=0
for i in range(x, x+10):
for j in range(y-10,y+10):
if pic[i,j]!=0:
c=c+1
avgi=avgi+i
avgj=avgj+j
recpx.append(i)
recpy.append(j)
if c==0:
break
newx=(avgi//c)+1
newy=(avgj//c)
if newx+10>120:
break
x=newx
y=newy
return recpx, recpy, colourpic
#sliding window down to up
def slidingwindowdup( x, y, pic, colourpic):
recpx=[]
recpy=[]
x=int(x)
y=int(y)
while(True):
cv2.rectangle(colourpic,(y-10,x-10),(y+10,x),(0,255,0),1)
avgi=0
avgj=0
c=0
for i in range(x-10,x):
for j in range(y-10,y+10):
if pic[i,j]!=0:
c=c+1
avgi=avgi+i
avgj=avgj+j
recpx.append(i)
recpy.append(j)
if c==0:
break
newx=(avgi//c)-1
newy=(avgj//c)
x=newx
y=newy
return recpx, recpy, colourpic
model=load_model('/home/ghosh/Documents/full_CNN_model.h5')
cap = cv2.VideoCapture('/home/ghosh/Documents/videoplayback3.mp4')
recent_fit=[]
avg_fit=[]
while(True):
ret, frame = cap.read()
small_img = imresize(frame, (80,160,3))
small_img = np.array(small_img)
small_img = small_img[None,:,:,:]
pred=model.predict(small_img)[0] *255
recent_fit.append(pred)
if len(recent_fit)>5:
recent_fit= recent_fit[1:]
avg_fit=np.mean(np.array([i for i in recent_fit]), axis=0)
blanks=np.zeros_like(avg_fit).astype(np.uint8)
lane_drawn=np.dstack((blanks, avg_fit, blanks))
imagef=imresize(frame,(720,1280,3))
lane_image=imresize(lane_drawn, (720,1280,3))
result=cv2.addWeighted(imagef,1,lane_image,1,0)
#****edge detection****#
img=cv2.cvtColor(lane_drawn, cv2.COLOR_BGR2GRAY)
img=np.uint8(img)
img=cv2.resize(img,(480,240))
blur = cv2.GaussianBlur(img,(5,5),0)
canny=cv2.Canny(blur,100,200)
erosion = cv2.erode(canny,(5,5),iterations = 1)
dilation = cv2.dilate(erosion,(5,5),iterations = 3)
roi=dilation[120:240,0:480]
#cv2.imshow("roi",roi)
pts1 = np.float32([[10,120],[450,120],[0,0],[480,0]])
pts2 = np.float32([[175,120],[230,120],[0,0],[480,0]])
M = cv2.getPerspectiveTransform(pts1,pts2)
dst = cv2.warpPerspective(roi,M,(480,120))
dst = dst.astype(np.uint8)
#cv2.imshow("transform",dst)
#print('image dtype ',dst.dtype)
#****histogram****#
c=0
wpixel=[]
rows, columns=dst.shape
#print("rows="+str(rows))
#print("columns="+str(columns))
for j in range(0,columns):
for i in range(0, rows):
if dst[i,j]!=0:
#print("["+str(i)+","+str(j)+"]="+str(dst[i,j]))
c=c+1
wpixel.append(c)
c=0
#plt.subplot(211),plt.imshow(dst)
#plt.subplot(212),plt.plot(wpixel)
#plt.show()
#****sliding window search****#
ly,lx,ry,rx = startpixels(wpixel, dst)
#print("Left lane start point="+str(lx)+","+str(ly))
#print("Right lane start point="+str(rx)+","+str(ry))
#print()
colourpic1= cv2.cvtColor(dst,cv2.COLOR_GRAY2RGB)
colourpic2= cv2.cvtColor(dst,cv2.COLOR_GRAY2RGB)
#left lane pixels
if lx<100:
leftlanex, leftlaney, sw1pic= slidingwindowdown( lx, ly, dst, colourpic1)
else:
leftlanex, leftlaney, sw1pic= slidingwindowdup( lx, ly, dst, colourpic1)
#print("left lane pixels: ",leftlanex,leftlaney)
#print()
#right lane pixels
if rx<100:
rightlanex, rightlaney, sw2pic= slidingwindowdown( rx, ry, dst, sw1pic)
else:
rightlanex, rightlaney, sw2pic= slidingwindowdup( rx, ry, dst, sw1pic)
#print("right lane pixels: ",rightlanex,rightlaney)
#print()
cv2.imshow("sliding window",sw2pic)
#****quadratic curve****#
leftz = np.polyfit( leftlanex,leftlaney, 2)
rightz = np.polyfit( rightlanex,rightlaney, 2)
funcl=np.poly1d(leftz)
funcr=np.poly1d(rightz)
#print("left lane polynomials:",leftz)
#print("right lane polynomials:",rightz)
#print(funcl)
#print(funcr)
xl_new = np.linspace(leftlanex[0], leftlanex[-1], 50)
yl_new = funcl(xl_new)
xr_new = np.linspace(rightlanex[0], rightlanex[-1], 50)
yr_new = funcr(xr_new)
lfp=[]
rfp=[]
for i in range(0,50):
lfp.append([yl_new[i],xl_new[i]])
for i in range(0,50):
rfp.append([yr_new[i],xr_new[i]])
lfpts=np.asarray(lfp)
rfpts=np.asarray(rfp)
#print(lfpts)
#print(rfpts)
cv2.polylines(colourpic2, np.int32([lfpts]),False, (255,0,0),1)
cv2.polylines(colourpic2, np.int32([rfpts]),False, (255,0,0),1)
#cv2.imshow("curve function",colourpic2)
#plt.imshow(colourpic2)
#plt.show()
#****radius of curvature****#
xm_per_pix = 27/45
ym_per_pix = 3.7/35
leftlanex = list( map(lambda x: x*xm_per_pix, leftlanex) )
leftlaney = list( map(lambda y: y*ym_per_pix, leftlaney) )
rightlanex = list( map(lambda x: x*xm_per_pix, rightlanex) )
rightlaney = list( map(lambda y: y*ym_per_pix, rightlaney) )
poly_coef_l = np.polyfit(leftlanex , leftlaney , 2)
radius_l = ((1 + (2 * poly_coef_l[0] * 45 * xm_per_pix + poly_coef_l[1]) ** 2) ** 1.5) / np.absolute(2 * poly_coef_l[0])
poly_coef_r = np.polyfit(rightlanex , rightlaney , 2)
radius_r = ((1 + (2 * poly_coef_r[0] * 45 * xm_per_pix + poly_coef_r[1]) ** 2) ** 1.5) / np.absolute(2 * poly_coef_r[0])
#print("left lane radius: ",radius_l)
#print("right lane radius: ",radius_r)
cv2.putText(result,'Left Lane Radius:'+str(radius_l),(30,40), cv2.FONT_HERSHEY_SIMPLEX, 1,(255,255,255),2,cv2.LINE_AA)
cv2.putText(result,'Right Lane Radius:'+str(radius_r),(30,80), cv2.FONT_HERSHEY_SIMPLEX, 1, (255,255,255),2,cv2.LINE_AA)
center= dst.shape[0]/2
rightx=funcr(5)
leftx=funcl(5)
offset=((rightx-center)-(center-leftx))*(ym_per_pix)
if offset<0:
cv2.putText(result,'Offset:'+str(np.abs(offset))+'m right',(30,120), cv2.FONT_HERSHEY_SIMPLEX, 1,(255,255,255),2,cv2.LINE_AA)
elif offset>0:
cv2.putText(result,'Offset:'+str(np.abs(offset))+'m left',(30,120), cv2.FONT_HERSHEY_SIMPLEX, 1,(255,255,255),2,cv2.LINE_AA)
else:
cv2.putText(result,'Offset:'+str(np.abs(offset))+'m',(30,120), cv2.FONT_HERSHEY_SIMPLEX, 1,(255,255,255),2,cv2.LINE_AA)
cv2.imshow('output',result)
#plt.subplot(221),plt.imshow(img),plt.title('Input')
#plt.subplot(222),plt.imshow(erosion),plt.title('Edge Detection')
#plt.subplot(223),plt.imshow(roi),plt.title('Roi')
#plt.subplot(224),plt.imshow(dst),plt.title('Perspective Transform')
#plt.show()
if cv2.waitKey(1) & 0xFF== ord('q'):
cv2.destroyAllWindows()
cap.release()