forked from visualbuffer/copilot
-
Notifications
You must be signed in to change notification settings - Fork 0
/
lane_detection.py
753 lines (636 loc) · 32 KB
/
lane_detection.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
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
import cv2
import numpy as np
import math
from datetime import datetime
from matplotlib import pyplot as plt
from collections import deque
from scipy.stats import mode
from scipy.optimize import curve_fit
from yolo_model import BoundBox
temp_dir = "images/detection/detect.jpg"
WHITE = (255, 255, 255)
YELLOW = (66, 244, 238)
GREEN = (80, 220, 60)
LIGHT_CYAN = (255, 255, 224)
DARK_BLUE = (139, 0, 0)
GRAY = (128, 128, 128)
BLUE = (255,0,0)
RED = (0,0,255)
ORANGE =(0,165,255)
BLACK =(0,0,0)
vehicles = [1,2,3,5,6,7,8]
animals =[15,16,17,18,19,21,22,23,]
humans =[0]
obstructions = humans + animals + vehicles
classes = [#
'Ped','bicycle','car','motorbike','aeroplane','bus',\
'train','truck','boat','traffic light','fire hydrant','stop sign',\
'parking meter','bench','bird','cat','dog','horse',\
'sheep','cow','elephant', 'bear','zebra','giraffe',\
'backpack','umbrella','handbag','tie','suitcase','frisbee',\
'skis','snowboard','sports ball','kite','baseball bat',\
'baseball glove','skateboard','surfboard','tennis racket','bottle','wine glass',\
'cup','fork','knife','spoon','bowl','banana',\
'apple','sandwich','orange','broccoli','carrot','hot dog',\
'pizza','donut','cake','chair','sofa','pottedplant',\
'bed','diningtable','toilet','tvmonitor','laptop','mouse',\
'remote','keyboard','cell phone','microwave','oven','toaster',\
'sink','refrigerator','book','clock','vase','scissors',\
'teddy bear','hair drier','toothbrush' ]
def create_queue(length = 10):
return deque(maxlen=length)
def polyfunc(x, a2, a1, a0):
return a2*x*x + a1*x + a0
class OBSTACLE(BoundBox):
xmax :int
xmin :int
ymin :int
ymax :int
xmid :int
ymid :int
lane : str
x : int
y : int
tracker = None
position : [int,int]
PERIOD = 4
__count = 0
def __init__(self,box: BoundBox, _id, v_updt =5) :
self.col_time:float =999.0
self._id = _id
self.position_hist = create_queue(v_updt)
# self.position_hist.append(dst)
self.update_coord(box)
self.update_score(box)
self.velocity = np.zeros((2))
self.score=box.score
self.label = box.label
def update_obstacle(self, dst, fps, n=5) :
self.position = dst
if self.lane == "my" :
self.col_time = min(int(dst[1]/(self.velocity[1]+0.001)*18/5),99)
else :
self.col_time = None
if (self.__count > self.position_hist.maxlen):
self.velocity = ((self.position-self.position_hist[0] ) * fps / self.position_hist.maxlen *5/18 ).astype(int)
self.__count += 1
self.position_hist.append(dst)
def update_coord(self,box):
self.xmax = box.xmax
self.xmin = box.xmin
self.ymin = box.ymin
self.ymax = box.ymax
self.xmid = int((box.xmax+box.xmin)/2)
self.ymid = int((box.ymax+box.ymin)/2)
self.position = np.mean(self.position_hist, axis = 0)
def update_score(self,box):
self.score=box.score
self.label = box.label
def update_box(self,box):
self.update_coord(box)
self.update_score(box)
class LANE_HISTORY:
def __init__(self,fps, queue_depth=12,poly_col=np.array([1,1,1]), test_points=[300, 500, 700], poly_max_deviation_distance=50, smoothing = 10, ploty =np.array([])):
self.fps =fps
self.test_points = np.asarray(test_points)
self.poly_max_deviation_distance = poly_max_deviation_distance
self.lost = False
self.max_lost_count = queue_depth
self.lost_count = self.max_lost_count + 10
self.smoothing = smoothing
self.ploty =ploty
self.leftFit = None # LEFT FIT POINTS
self.rightFit = None # RIGHT FIT POINTS
self.leftx = create_queue(self.fps//4)
self.rightx = create_queue(self.fps//4)
self.width = None
self.previous_centers = None
self.current_coef = None
self.smoothed_poly = poly_col
self.poly_history = create_queue(queue_depth)
self.y = None
self.x = None
self.appended = 0
self.breached = 0
self.reset = 0
self.curvature = 0
self.centerx = 0
self.lane_offset = 0
self.left_windows = []
self.right_windows=[]
def compute_lane_points(self) :
self.leftFit = self.previous_centers - self.width//2
self.rightFit = self.previous_centers + self.width//2
def compute_curvature(self, alpha, beta):
y_eval = -np.max(self.ploty)
lp = self.smoothed_poly
self.curvature = int(((beta**2 + (2 * lp[0] * y_eval * alpha**2 + \
lp[1]*alpha)**2)**1.5)/(np.absolute(2 * lp[0]*(alpha*beta)**2)))
return
def compute_offset(self):
y_eval = -np.max(self.ploty)
lp = self.smoothed_poly
self.lane_offset = lp[0] * y_eval**2 + lp[1] * y_eval + lp[2] - self.centerx
if abs(self.lane_offset) > self.width //2 :
if self.lane_offset < 0 :
print("\n\rLANE CHANGE TO RIGHT\033[F")
self.poly_history = create_queue(self.poly_history.maxlen)
self.leftx = self.rightx
self.rightx = create_queue(length = self.rightx.maxlen)
self.rightx.append(int(np.mean(self.leftx) + self.width ))
self.previous_centers = self.previous_centers+self.width
else :
print("\n\rLANE CHANGE TO LEFT\033[F")
self.poly_history = create_queue(self.poly_history.maxlen)
self.rightx = self.leftx
self.leftx = create_queue(length = self.leftx.maxlen)
self.leftx.append(int(np.mean(self.rightx) - self.width ))
self.previous_centers = self.previous_centers-self.width
else:
self.leftx.append(self.previous_centers[0] - self.width//2)
self.rightx.append(self.previous_centers[0] + self.width//2)
return
def addlane(self, y,x):
status = "APPENDED | "
self.y = y
self.x = x
self.current_coef,_ = curve_fit(polyfunc,self.y, self.x, p0=self.smoothed_poly)
if (self.lost_count > self.max_lost_count ) :
status ="RESET | "
self.get_smoothed_polynomial()
self.lost = False
self.lost_count = 0
self.reset +=1
return True, status
test_y_smooth = np.asarray(list(map(lambda x: np.polyval(self.smoothed_poly,x), -self.test_points)))
test_y_new = np.asarray(list(map(lambda x: np.polyval(self.current_coef,x), -self.test_points)))
dist = np.absolute(test_y_smooth - test_y_new)
max_dist = dist[np.argmax(dist)]
if max_dist > self.poly_max_deviation_distance:
status = "BREACHED | "
self.lost = True
self.lost_count += 1
self.breached +=1
return False , status
self.get_smoothed_polynomial()
self.lost = False
self.lost_count = 0
self.appended +=1
return True, status
def get_smoothed_polynomial(self):
self.poly_history.append(self.current_coef)
all_coeffs = np.asarray(list(self.poly_history))
self.smoothed_poly = np.mean(all_coeffs[-self.smoothing:,:], axis=0)
self.previous_centers = np.asarray([np.polyval(self.smoothed_poly,-x) for x in self.ploty], dtype=int)
self.compute_lane_points()
self.compute_offset()
return self.smoothed_poly
def calculate_position(self,x,y):
position = np.polyval(self.smoothed_poly ,y) - x
status = "right"
if position < - self.width//2 :
status = "left"
elif position < self.width//2:
status = "my"
return status
class LANE_DETECTION:
"""
The AdvancedLaneDetectorWithMemory is a class that can detect lines on the road
"""
UNWARPED_SIZE :(int,int)
WRAPPED_WIDTH : int
_pip_size=(int,int)
_pip__x_offset=20
_pip__y_offset=10
img_dimensions=(int,int)
temp_dir = "./images/detection/"
windows_per_line = 30
vanishing_point:(int,int)
real_world_lane_size_meters=(32, 3.7)
font = cv2.FONT_HERSHEY_SIMPLEX
bottom = 0
def __init__(self, img,fps,
yellow_lower = np.uint8([ 20, 50, 50]),
yellow_upper = np.uint8([35, 255, 255]),
white_lower = np.uint8([ 0, 200, 0]),
white_upper = np.uint8([180, 255, 100]),
lum_factor = 130,
max_gap_th = 2/5,
lane_start=[0.35,0.75] ,
verbose = 3):
self.message = ""
self.verbose = verbose
self.objpts = None
self.count = 0
self.fps=int(fps)
self.imgpts = None
self.lane_roi = None
## LANE DETECTION PROPERTIES
self.yellow_lower = yellow_lower
self.yellow_upper = yellow_upper
self.white_lower = white_lower
self.white_upper = white_upper
self.lane_change = False
# self.lane_width = 0
# IMAGE PROPERTIES
self.lum_factor = lum_factor
self.image = img
self.font_sz = 4e-4 * self.image.shape[0]
self.img_dimensions = (self.image.shape[0], self.image.shape[1])
self.UNWARPED_SIZE = (360,360)#(int(self.img_dimensions[1]*0.5),int(self.img_dimensions[1]*0.5))
self.WRAPPED_WIDTH = int(self.img_dimensions[1]*0.15)
self.margin = int(self.UNWARPED_SIZE[1]*0.08)
self.window_height = np.int(self.UNWARPED_SIZE[1]//self.windows_per_line)
self._pip_size = (int(self.image.shape[1] * 0.2), int(self.image.shape[0]*0.2))
self.minpix=self.window_height
self.maxpix = int(self.margin * self.window_height *0.5)
self.n_gap_skip = 0
self.max_gap = 0
self.windows_range = range( self.windows_per_line)
self.window_offset = 0
self.ploty = np.linspace(int(self.UNWARPED_SIZE[1]*0.45), self.UNWARPED_SIZE[1]- 1, int(self.UNWARPED_SIZE[1]*0.4), dtype=int)
test = np.arange(0.3,1,0.1)*self.UNWARPED_SIZE[1]
test = test.astype(int)
self.lane = LANE_HISTORY(self.fps,test_points = test, queue_depth=self.fps//3, ploty = self.ploty)
self.max_gap_th = max_gap_th * self.windows_per_line
self.calc_perspective(lane_start=lane_start)
def compute_bounds(self, image):
lx = int( max( np.mean(self.lane.leftx),0))
rx = int(max(min(np.mean(self.lane.rightx), image.shape[0]),max(image.shape[0]//4,lx)))
avg = np.average(image[lx:rx,\
image.shape[1]*2//3 : image.shape[1]-self.bottom,1])
l_rel = max(min((avg/self.lum_factor)**2,1.3),0.45)
self.yellow_lower[1] = int(l_rel*30)
self.white_lower[1] = int(l_rel *170)
self.message = self.message+"WHITE" + str(self.white_lower[1])
return
def compute_mask(self, image):
"""
Returns a binary thresholded image produced retaining only white and yellow elements on the picture
The provided image should be in RGB formats
"""
# cv2.imwrite(self.temp_dir+"temp.jpg", image)
converted = cv2.cvtColor(image, cv2.COLOR_BGR2HLS)
if self.count % (self.fps*4) == 0 :
self.compute_bounds(converted)
yellow_mask = cv2.inRange(converted, self.yellow_lower, self.yellow_upper)
white_mask = cv2.inRange(converted, self.white_lower, self.white_upper)
mask = cv2.bitwise_or(white_mask, yellow_mask)
# t2 = cv2.bitwise_and(image, image, mask = mask)
# cv2.imwrite(self.temp_dir+"temp2.jpg", t2)
# print(self.white_lower[1])
return mask
def calc_perspective(self, lane_start=[0.25,0.75]):
roi = np.zeros((self.img_dimensions[0], self.img_dimensions[1]), dtype=np.uint8) # 720 , 1280
roi_points = np.array([[0, self.img_dimensions[0]*7//9],
[0, self.img_dimensions[0]],
[self.img_dimensions[1], self.img_dimensions[0]],
[self.img_dimensions[1], self.img_dimensions[0]*7//9],
[self.img_dimensions[1]*45//99,self.img_dimensions[0]//2],
[self.img_dimensions[1]*45//99,self.img_dimensions[0]//2]], dtype=np.int32)
cv2.fillPoly(roi, [roi_points], 1)
self.lane_roi = np.zeros((self.img_dimensions[0], self.img_dimensions[1]), dtype=np.uint8)
Lhs = np.zeros((2,2), dtype= np.float32)
Rhs = np.zeros((2,1), dtype= np.float32)
grey = cv2.cvtColor(self.image, cv2.COLOR_BGR2GRAY)
mn_hsl = np.median(grey)
edges = cv2.Canny(grey, int(mn_hsl), int(mn_hsl*.3))
lines = cv2.HoughLinesP(edges*roi,rho =self.img_dimensions[0]//20,\
theta = 2* np.pi/180,\
threshold = self.img_dimensions[0]//80,\
minLineLength = self.img_dimensions[0]//3,\
maxLineGap = self.img_dimensions[0]//15)
img2 = self.image.copy()
for line in lines:
for x1, y1, x2, y2 in line:
cv2.line(img2,(x1,y1),(x2,y2),(255,0,0),2)
normal = np.array([[-(y2-y1)], [x2-x1]], dtype=np.float32)
normal /=np.linalg.norm(normal)
point = np.array([[x1],[y1]], dtype=np.float32)
outer = np.matmul(normal, normal.T)
Lhs += outer
Rhs += np.matmul(outer, point)
self.vanishing_point = np.matmul(np.linalg.inv(Lhs),Rhs).reshape(2)
orig_points=self.vanishing_point.copy()
if abs(self.vanishing_point[0] - self.img_dimensions[1]//2) > 0.07 *self.img_dimensions[1] :
print("ABSURD X POSITION TRY OTHER PARAMETERS",self.vanishing_point[0] , "in ==> ",self.img_dimensions )
self.vanishing_point[0] = self.img_dimensions[1]//2
if abs(self.vanishing_point[1] - self.img_dimensions[0]*0.57) > 0.1 *self.img_dimensions[0] :
print("ABSURD Y POSITION TRY OTHER PARAMETERS",self.vanishing_point[0] , "in ==>",self.img_dimensions )
self.vanishing_point[1] = int(self.img_dimensions[0]*0.57)
top =self.vanishing_point[1] + int(self.WRAPPED_WIDTH*0.15)
self.bottom = int(0.02*self.img_dimensions[0])
bottom = self.img_dimensions[0]+self.bottom
def on_line(p1, p2, ycoord):
return [p1[0]+ (p2[0]-p1[0])/float(p2[1]-p1[1])*(ycoord-p1[1]), ycoord]
p1 = [self.vanishing_point[0] - self.WRAPPED_WIDTH/2, top]
p2 = [self.vanishing_point[0] + self.WRAPPED_WIDTH/2, top]
p3 = on_line(p2,self.vanishing_point, bottom)
p4 = on_line(p1,self.vanishing_point, bottom)
src_points = np.array([p1,p2,p3,p4], dtype=np.float32)
dst_points = np.array([[0, 0], [self.UNWARPED_SIZE[0], 0],
[self.UNWARPED_SIZE[0], self.UNWARPED_SIZE[1]],
[0, self.UNWARPED_SIZE[1]]], dtype=np.float32)
self.trans_mat = cv2.getPerspectiveTransform(src_points, dst_points)
self.inv_trans_mat = cv2.getPerspectiveTransform(dst_points,src_points)
min_wid = 1000
img = cv2.warpPerspective(self.image, self.trans_mat, self.UNWARPED_SIZE)
x1 = int(self.UNWARPED_SIZE[0]*lane_start[0])
x2 = int(self.UNWARPED_SIZE[0]*lane_start[1])
self.lane.leftx.append(x1)
self.lane.rightx.append(x2)
mask = self.compute_mask(img)
# x = np.linspace(0,mask.shape[0]-1,mask.shape[0])
span = self.UNWARPED_SIZE[0]//5
x1 = x1-span + self.detect_lane_start(mask[:,x1-span :x1+span])
x2 = x2-span + self.detect_lane_start(mask[:,x2-span :x2+span])
self.lane.leftx.append(x1)
self.lane.rightx.append(x2)
self.lane.width = x2 -x1
self.lane.previous_centers = np.ones(self.windows_per_line)*(x1+x2)//2
lane_roi_points = np.array([
[self.img_dimensions[1]*9//80, self.img_dimensions[0]],
[self.img_dimensions[1]*71//80,self.img_dimensions[0]],
[self.vanishing_point[0] + 3*self.img_dimensions[1]//25,self.vanishing_point[1] - 10],
[self.vanishing_point[0] - 3*self.img_dimensions[1]//25,self.vanishing_point[1] - 10]], dtype=np.int32)
cv2.fillPoly(self.lane_roi , [lane_roi_points], 1)
# self.lane_roi = self.lane_roi*grad
if (x2-x1<min_wid):
min_wid = x2-x1
self.px_per_xm = min_wid/self.real_world_lane_size_meters[1]
self.xm_per_px = 1/self.px_per_xm
if False :#self.camera.callibration_done :
Lh = 1#np.linalg.inv(np.matmul(self.trans_mat, self.camera.cam_matrix))
else:
Lh = np.linalg.inv(self.trans_mat)
self.px_per_ym = self.px_per_xm * np.linalg.norm(Lh[:,0]) / np.linalg.norm(Lh[:,1])
self.ym_per_px = 1/self.px_per_ym
self.perspective_done_at = datetime.utcnow().timestamp()
pos = np.array((self.vanishing_point[0], bottom )).reshape(1, 1, -1)
dst = cv2.perspectiveTransform(pos, self.trans_mat).reshape(2)
self.lane.centerx = dst[0]
print("PERSPECTIVE TRANSFORMATION MATRIX COMPUTED")
if self.verbose > 1:
img_orig = cv2.polylines(self.image, [src_points.astype(np.int32)],True, (0,0,255), thickness=5)
cv2.line(img, (int(x1), 0), (int(x1), self.UNWARPED_SIZE[1]), (255, 0, 0), 3)
cv2.line(img, (int(x2), 0), (int(x2), self.UNWARPED_SIZE[1]), (0, 0, 255), 3)
cv2.circle(img_orig,tuple(self.vanishing_point),10, color=RED, thickness=5)
cv2.circle(img_orig,tuple(orig_points),10, color=GRAY, thickness=4)
cv2.imwrite(self.temp_dir+"vanishing_point.jpg",img_orig)
cv2.imwrite(self.temp_dir+"lane_width.jpg",img)
cv2.imwrite(self.temp_dir+"perspective_lines.jpg",img2)
cv2.imwrite(self.temp_dir+"mask.jpg",mask)
img = cv2.bitwise_and(img, img, mask = mask )
cv2.imwrite(self.temp_dir+"masked_regions.jpg",img)
cv2.imwrite(self.temp_dir+"edges.jpg",edges*roi)
return img_orig
return
def calculate_position(self, box: OBSTACLE):
pos = np.array((box.xmax/2+box.xmin/2, box.ymax)).reshape(1, 1, -1)
dst = cv2.perspectiveTransform(pos, self.trans_mat).reshape(2)
box.x = int(dst[0])
box.y = -int(dst[1])
box.lane = self.lane.calculate_position(box.x,box.y)
dst = np.array([dst[0]/self.px_per_xm,(self.UNWARPED_SIZE[1]-dst[1])/self.px_per_ym])
box.update_obstacle(dst,self.fps)
return box
# else:
# return np.array([0,0])
def put_text(self, overlay,text, coord, color=WHITE):
sz = self.font_sz*50
rect_ht = int(sz *1.1)
rect_wd = int(len(text)*sz*0.5)
p1 = (coord[0], coord[1]+2)
p2 = (coord[0]+rect_wd, coord[1]-rect_ht)
cv2.rectangle(overlay, p1, p2, (0, 0, 0),-1)
cv2.putText(overlay, text, coord, self.font, self.font_sz*1.25, color, 1, cv2.LINE_AA)
return
def process_image(self, img,obstacles :[OBSTACLE] =[], alpha=.7, beta=0.3, gamma=0):
"""
Attempts to find lane lines on the given image and returns an image with lane area colored in green
as well as small intermediate images overlaid on top to understand how the algorithm is performing
"""
off = int(100*self.font_sz)
self.message = "["+ str(self.count)+"]"
overlay = img.copy()
thres_img_psp= self.compute_lane_lines(overlay)
for i in range(len(obstacles)):
obstacles[i] = self.calculate_position(obstacles[i])
box = obstacles[i]
past=[box.xmin,box.ymin,box.xmax,box.ymax]
color = WHITE
t1 = classes[obstructions[box.label]] +" ["+str(int(box.position[1])) + "m]"
t2 = "("+str(int(box.score*100))+"%) ID: " +str(box._id)
b1= "Lane " + box.lane + " " + str(int(box.velocity[1]))+"kmph"
pt1 = (box.xmin, box.ymin-off)
pt2 = (box.xmin, box.ymin)
pb1 = (box.xmin, box.ymax+off)
if (box.lane == "my") and (box.velocity[1] < 0) :
color = RED
if box.col_time > -4 :
b3 = "Col "+str(int(box.col_time))+"s"
pb3 = (box.xmin, box.ymax+2*off)
self.put_text(overlay, b3, pb3, color = color)
self.put_text(overlay, t1, pt1, color = color)
self.put_text(overlay, t2, pt2, color = color)
self.put_text(overlay, b1, pb1, color = color)
past_center = (int(past[0]/2+past[2]/2), past[3])
color = ORANGE if box.velocity[1] < 0 else GREEN
cv2.rectangle(overlay, (box.xmin,box.ymin), (box.xmax,box.ymax), color,2)
cv2.circle(overlay,past_center,1, GRAY,2)
img = cv2.addWeighted(img, alpha, overlay, beta, gamma)
out_img = np.dstack((thres_img_psp,thres_img_psp,thres_img_psp))
img = self.draw_lane_area(out_img, img)
if self.verbose >2 :
drawn_lines = self.draw_lane_lines(out_img)
drawn_hotspots = self.draw_lines_hotspots(out_img, obstacles)
img = self.combine_images(img, drawn_lines,drawn_hotspots)
img = self.draw_lane_curvature_text(img,)
return img
def draw_lane_curvature_text(self, img):
"""
Returns an image with curvature information inscribed
"""
sz = self.font_sz*3
offset_y = self._pip_size[1] * 1 + self._pip__y_offset * 5
offset_x = self._pip__x_offset
template = "{0:17}{1:17}"
txt_header = template.format("Curvature ", "Offset")
# print(txt_header)
txt_values = template.format("{:d}m".format(self.lane.curvature),
"{:.2f}m Left".format(self.lane.lane_offset*self.xm_per_px))
if self.lane.lane_offset < 0.0:
txt_values = template.format("{:d}m".format(self.lane.curvature),
"{:.2f}m Right".format(self.lane.lane_offset*self.xm_per_px))
cv2.putText(img, txt_header, (offset_x, offset_y), self.font, sz, BLACK, 1, cv2.LINE_AA)
cv2.putText(img, txt_values, (offset_x, offset_y + self._pip__y_offset * 2), self.font, sz, BLACK, 2, cv2.LINE_AA)
cv2.putText(img, self.message, (offset_x, self.img_dimensions[0]-10), self.font, sz, BLACK, 1, cv2.LINE_AA)
return img
def combine_images(self, lane_area_img, lines_img,lane_hotspots_img):
"""
Returns a new image made up of the lane area image, and the remaining lane images are overlaid as
small images in a row at the top of the the new image
"""
small_lines = cv2.resize(lines_img, self._pip_size)
small_hotspots = cv2.resize(lane_hotspots_img, self._pip_size)
lane_area_img[self._pip__y_offset: self._pip__y_offset + self._pip_size[1], self._pip__x_offset: self._pip__x_offset + self._pip_size[0]] = small_lines
start_offset_y = self._pip__y_offset
start_offset_x = 2 * self._pip__x_offset + 1 * self._pip_size[0]
lane_area_img[start_offset_y: start_offset_y + self._pip_size[1], start_offset_x: start_offset_x + self._pip_size[0]] = small_hotspots
return lane_area_img
def draw_lane_area(self, warped_img, undist_img):
"""
Returns an image where the inside of the lane has been colored in bright green
"""
color_warp = np.zeros_like(warped_img).astype(np.uint8)
pts_left = np.array([np.transpose(np.vstack([self.lane.leftFit, self.ploty]))])
pts_right = np.array([np.flipud(np.transpose(np.vstack([self.lane.rightFit, self.ploty])))])
pts = np.hstack((pts_left, pts_right))
cv2.fillPoly(color_warp, np.int_([pts]),GREEN)
newwarp = cv2.warpPerspective(color_warp, self.inv_trans_mat, (undist_img.shape[1], undist_img.shape[0]))
result = cv2.addWeighted(undist_img, 1, newwarp, 0.3, 0)
return result
def draw_lane_lines(self, img):
"""
Returns an image where the computed lane lines have been drawn on top of the original warped binary image
"""
out_img = img.copy()
for low_pt, high_pt in self.lane.left_windows:
cv2.rectangle(out_img, low_pt, high_pt, (0, 255, 0), 1)
for low_pt, high_pt in self.lane.right_windows:
cv2.rectangle(out_img, low_pt, high_pt, (0, 255, 0), 1)
return out_img
def draw_lines_hotspots(self, img, obstacles:[OBSTACLE] = []):
"""
Returns a RGB image where the portions of the lane lines that were
identified by our pipeline are colored in yellow (left) and blue (right)
"""
sz = self.font_sz*12
out_img = img.copy()
lx = self.lane.x - self.lane.width//2
rx = self.lane.x + self.lane.width//2
pts_left = np.dstack((self.lane.leftFit, self.ploty)).astype(np.int32)
pts_right = np.dstack((self.lane.rightFit, self.ploty)).astype(np.int32)
pts_cntr = np.dstack((self.lane.rightFit - self.lane.width//2, self.ploty)).astype(np.int32)
cv2.polylines(out_img, pts_left, False, BLUE, 2)
cv2.polylines(out_img, pts_right, False, RED, 2)
cv2.polylines(out_img, pts_cntr, False, YELLOW, 15)
for i in range(len(self.lane.x)) :
cv2.circle( out_img,( lx[i],-self.lane.y[i]), 4, BLUE, -1)
cv2.circle( out_img,( rx[i],-self.lane.y[i]), 4, RED, -1)
for i in range(len(obstacles)):
box = obstacles[i]
color = GREEN
if (box.col_time) and (box.lane == "my") and (box.col_time < 0) and (box.col_time>-4) :
color = RED
cv2.putText(out_img,str(box._id),(box.x, -box.y), self.font, sz, color, 8, cv2.LINE_AA)
return out_img
def detect_lane_start(self, image):
histx = np.sum(image[image.shape[0]*4//5:,:], axis=0)
return np.argmax(histx)
def compute_lane_lines(self, img):
"""
Returns the tuple (left_lane_line, right_lane_line) which represents respectively the LANE_LINE instances for
the computed left and right lanes, for the supplied binary warped image
"""
self.lane.left_windows = []
self.lane.right_windows = []
undst_img = cv2.bitwise_and(img, img, mask = self.lane_roi )
pp_img = cv2.warpPerspective(undst_img, self.trans_mat, (self.UNWARPED_SIZE[1],self.UNWARPED_SIZE[0]))
warped_img = self.compute_mask(pp_img)
x1_av = int(np.average(self.lane.leftx))
x2_av = int(np.average(self.lane.rightx))
self.lane.width= min(max(int(x2_av - x1_av), self.UNWARPED_SIZE[0]//3),self.UNWARPED_SIZE[0]//2)
x1 = min(max(x1_av,self.margin), self.UNWARPED_SIZE[0]-self.lane.width)
x2 = max( min(x2_av,self.UNWARPED_SIZE[0]-self.margin-1), self.lane.width)
leftx_current = x1-self.margin + self.detect_lane_start(warped_img[:,x1-self.margin :x1+self.margin])
rightx_current = x2-self.margin + self.detect_lane_start(warped_img[:,x2-self.margin :x2+self.margin])
nonzero = warped_img.nonzero()
nonzeroy = np.array(nonzero[0])
nonzerox = np.array(nonzero[1])
centerx_current = (x2_av - x1_av) //2
pointx = []
pointy=[]
center_idx = []
curve_compute = 0
self.max_gap = 0
gap = 0
for window in self.windows_range:
win_y_low = warped_img.shape[0] - (window + 1)* self.window_height
win_y_high = warped_img.shape[0] - window * self.window_height
win_xleft_low = leftx_current - self.margin
win_xleft_high = leftx_current + self.margin
win_xright_low = rightx_current - self.margin
win_xright_high = rightx_current + self.margin
self.lane.left_windows.append([(win_xleft_low,win_y_low),(win_xleft_high,win_y_high)])
good_left_inds = ((nonzeroy >= win_y_low) & (nonzeroy < win_y_high) &
(nonzerox >= win_xleft_low) & (nonzerox < win_xleft_high)).nonzero()[0]
self.lane.right_windows.append([(win_xright_low,win_y_low),(win_xright_high,win_y_high)])
good_right_inds = ((nonzeroy >= win_y_low) & (nonzeroy < win_y_high) &
(nonzerox >= win_xright_low) & (nonzerox < win_xright_high)).nonzero()[0]
centerx =np.array([])
centery =np.array([])
s = 0
if (len(good_left_inds) > self.minpix) and (len(good_left_inds) < self.maxpix):
x = mode(nonzerox[good_left_inds])[0]
y = mode(nonzeroy[good_left_inds])[0]
s +=1
centerx = x + self.lane.width//2
centery = -y
if (len(good_right_inds) > self.minpix) and (len(good_right_inds) < self.maxpix):
s += 1
x = mode(nonzerox[good_right_inds])[0]
y = mode(nonzeroy[good_right_inds])[0]
centerx = np.append(centerx, x-self.lane.width//2)
centery = np.append(centery,-y)
if s > 0:
centerx_current = int(np.average(centerx))
pointx.append(centerx_current)
pointy.append(int(np.average(centery)))
gap = 0
else :
gap +=1
if len(center_idx) > 5 :
if curve_compute% 5 == 0 :
self.coef,_ =curve_fit(polyfunc,pointy,np.array(pointx), p0=self.coef)
centerx_current = int(np.polyval(self.coef, (window+1)*self.window_height))
curve_compute += 1
else :
centerx_current = self.lane.previous_centers[window-self.window_offset]
self.max_gap = max(self.max_gap, gap)
leftx_current = int(centerx_current -self.lane.width/2)
rightx_current = int(centerx_current +self.lane.width/2)
if (not self.lane_change):
if (self.max_gap > self.max_gap_th) and (self.count>0) :
self.lane.left_windows = []
self.lane.right_windows = []
self.message+="SKIPPED "+ str(self.max_gap)
self.count+=1
self.n_gap_skip+=1
self.compute_bounds(cv2.cvtColor(pp_img, cv2.COLOR_BGR2HLS))
return warped_img
status , message = self.lane.addlane(pointy,np.array(pointx) )
self.message+= message
if not status :
self.compute_bounds(cv2.cvtColor(pp_img, cv2.COLOR_BGR2HLS))
else:
self.lane.compute_curvature(self.px_per_ym,self.px_per_xm)
self.count+=1
return warped_img
if __name__ == "__main__":
video_reader = cv2.VideoCapture("videos/nice_road.mp4")
fps = video_reader.get(cv2.CAP_PROP_FPS)
nb_frames = int(video_reader.get(cv2.CAP_PROP_FRAME_COUNT))
frame_h = int(video_reader.get(cv2.CAP_PROP_FRAME_HEIGHT))
frame_w = int(video_reader.get(cv2.CAP_PROP_FRAME_WIDTH))
pers_frame_time = 398#180# seconds
pers_frame = int(pers_frame_time *fps)
video_reader.set(1,pers_frame)
ret, image = video_reader.read()
ld = LANE_DETECTION( image,fps,
yellow_lower = np.uint8([ 20, 50, 110]),
yellow_upper = np.uint8([35, 255, 255]),
white_lower = np.uint8([ 0, 140, 0]),
white_upper = np.uint8([255, 255, 100]),
lum_factor = 110,
lane_start=[0.2,0.5])