Spaces:
Sleeping
Sleeping
File size: 36,729 Bytes
e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab 2c5dee6 e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab 2c5dee6 63ea6ab 2c5dee6 63ea6ab 2c5dee6 e40d5b1 63ea6ab 2c5dee6 63ea6ab e40d5b1 63ea6ab 2c5dee6 63ea6ab 2c5dee6 e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab 4ae3a76 63ea6ab 4ae3a76 63ea6ab 4ae3a76 63ea6ab 4ae3a76 63ea6ab 4ae3a76 63ea6ab 4ae3a76 e40d5b1 4ae3a76 63ea6ab 4ae3a76 e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 3e26e16 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab e40d5b1 63ea6ab 4ae3a76 63ea6ab e40d5b1 63ea6ab 4ae3a76 63ea6ab e40d5b1 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 |
"""
Lane detection module using OpenCV
Advanced lane detection with multiple methods and GPU acceleration.
This module contains the core lane detection logic without UI dependencies.
"""
import cv2
import numpy as np
# GPU acceleration setup - prioritize NVIDIA GPU
USE_GPU = False
GPU_TYPE = "none"
try:
if cv2.cuda.getCudaEnabledDeviceCount() > 0:
USE_GPU = True
GPU_TYPE = "nvidia"
cv2.cuda.setDevice(0) # Use first GPU
print(f"✓ NVIDIA CUDA enabled! Using GPU acceleration on device: {cv2.cuda.printShortCudaDeviceInfo(cv2.cuda.getDevice())}")
else:
print("ℹ CUDA not available. Using CPU.")
except Exception as e:
print(f"ℹ GPU acceleration not available: {e}. Using CPU.")
GPU_TYPE = "none"
def region_of_interest(img, vertices):
"""
Apply a region of interest mask to the image.
"""
mask = np.zeros_like(img)
cv2.fillPoly(mask, vertices, 255)
masked_image = cv2.bitwise_and(img, mask)
return masked_image
def draw_lines_basic(img, lines, color=[0, 255, 0], thickness=3):
"""
Draw lines on the image with filled lane area (Basic method).
- Lane lines: Red color
- Lane interior: Green semi-transparent fill
"""
if lines is None:
return img
line_img = np.zeros_like(img)
# Separate left and right lane lines
left_lines = []
right_lines = []
for line in lines:
x1, y1, x2, y2 = line[0]
if x2 == x1:
continue
slope = (y2 - y1) / (x2 - x1)
# Filter by slope to separate left and right lanes
if slope < -0.5: # Left lane (negative slope)
left_lines.append(line[0])
elif slope > 0.5: # Right lane (positive slope)
right_lines.append(line[0])
# Average lines for left and right lanes
def average_lines(lines, img_shape):
if len(lines) == 0:
return None
x_coords = []
y_coords = []
for line in lines:
x1, y1, x2, y2 = line
x_coords.extend([x1, x2])
y_coords.extend([y1, y2])
# Fit a polynomial to the points
poly = np.polyfit(y_coords, x_coords, 1)
# Calculate line endpoints
y1 = img_shape[0]
y2 = int(img_shape[0] * 0.6)
x1 = int(poly[0] * y1 + poly[1])
x2 = int(poly[0] * y2 + poly[1])
return [x1, y1, x2, y2]
# Draw averaged lines
left_line = average_lines(left_lines, img.shape)
right_line = average_lines(right_lines, img.shape)
# Fill the lane area with green color
if left_line is not None and right_line is not None:
# Create polygon points for the lane area
lane_polygon = np.array([[
(left_line[0], left_line[1]), # Left bottom
(left_line[2], left_line[3]), # Left top
(right_line[2], right_line[3]), # Right top
(right_line[0], right_line[1]) # Right bottom
]], dtype=np.int32)
# Fill the lane area with green (semi-transparent)
cv2.fillPoly(line_img, lane_polygon, (0, 255, 0))
# Draw the lane lines in red with thicker lines
if left_line is not None:
cv2.line(line_img, (left_line[0], left_line[1]), (left_line[2], left_line[3]),
(0, 0, 255), thickness * 2) # Red color (BGR format)
if right_line is not None:
cv2.line(line_img, (right_line[0], right_line[1]), (right_line[2], right_line[3]),
(0, 0, 255), thickness * 2) # Red color (BGR format)
# Blend with original image (make the overlay semi-transparent)
return cv2.addWeighted(img, 0.8, line_img, 0.5, 0)
def draw_lines_segmented(img, lines, color=[0, 255, 0], thickness=3):
"""
Draw multiple short line segments to represent curves.
Better curve representation with Hough Transform.
- Lane lines: Red segmented lines
- Lane interior: Green semi-transparent fill
"""
if lines is None:
return img
line_img = np.zeros_like(img)
fill_img = np.zeros_like(img)
# Separate left and right lane lines
left_lines = []
right_lines = []
for line in lines:
x1, y1, x2, y2 = line[0]
if x2 == x1:
continue
slope = (y2 - y1) / (x2 - x1)
# Filter by slope to separate left and right lanes
if slope < -0.5: # Left lane (negative slope)
left_lines.append(line[0])
elif slope > 0.5: # Right lane (positive slope)
right_lines.append(line[0])
# Extract left and right lane boundaries
left_x = []
left_y = []
right_x = []
right_y = []
for line in left_lines:
x1, y1, x2, y2 = line
left_x.extend([x1, x2])
left_y.extend([y1, y2])
for line in right_lines:
x1, y1, x2, y2 = line
right_x.extend([x1, x2])
right_y.extend([y1, y2])
# Initialize sorted lists
left_x_sorted = []
left_y_sorted = []
right_x_sorted = []
right_y_sorted = []
# Sort by y coordinate to maintain order
if len(left_x) > 0:
left_coords = sorted(zip(left_y, left_x))
left_y_sorted = [c[0] for c in left_coords]
left_x_sorted = [c[1] for c in left_coords]
# Draw all individual line segments for left lane
for line in left_lines:
x1, y1, x2, y2 = line
cv2.line(line_img, (x1, y1), (x2, y2), (0, 0, 255), thickness + 2)
if len(right_x) > 0:
right_coords = sorted(zip(right_y, right_x))
right_y_sorted = [c[0] for c in right_coords]
right_x_sorted = [c[1] for c in right_coords]
# Draw all individual line segments for right lane
for line in right_lines:
x1, y1, x2, y2 = line
cv2.line(line_img, (x1, y1), (x2, y2), (0, 0, 255), thickness + 2)
# Fill the area between left and right lanes
if len(left_y_sorted) > 0 and len(right_y_sorted) > 0:
# Create a polygon by combining left and right points
min_y = max(min(left_y_sorted), min(right_y_sorted))
max_y = min(max(left_y_sorted), max(right_y_sorted))
if max_y > min_y:
# Interpolate to get matching y-coordinates
y_range = np.arange(int(min_y), int(max_y), 10)
poly_points = []
# Left points
for y in y_range:
if y >= min(left_y_sorted) and y <= max(left_y_sorted):
idx = np.searchsorted(left_y_sorted, y)
if idx > 0 and idx < len(left_x_sorted):
x = left_x_sorted[idx]
poly_points.append([x, y])
# Right points (reverse order for polygon)
for y in reversed(y_range):
if y >= min(right_y_sorted) and y <= max(right_y_sorted):
idx = np.searchsorted(right_y_sorted, y)
if idx > 0 and idx < len(right_x_sorted):
x = right_x_sorted[idx]
poly_points.append([x, y])
if len(poly_points) >= 3:
poly_points = np.array(poly_points, dtype=np.int32)
cv2.fillPoly(fill_img, [poly_points], (0, 255, 0))
# Combine filled area and lines
result_img = cv2.addWeighted(line_img, 0.6, fill_img, 0.7, 0)
# Blend with original image
return cv2.addWeighted(img, 0.8, result_img, 0.5, 0)
def process_frame_basic(frame, use_segmented=False):
"""
Process a single frame for lane detection using basic Hough Transform method.
use_segmented: If True, draw multiple line segments for better curve representation.
If False, draw averaged single line (default).
"""
height, width = frame.shape[:2]
if USE_GPU and GPU_TYPE == "nvidia":
# Upload frame to GPU
gpu_frame = cv2.cuda_GpuMat()
gpu_frame.upload(frame)
# Convert to grayscale on GPU
gpu_gray = cv2.cuda.cvtColor(gpu_frame, cv2.COLOR_BGR2GRAY)
# Apply Gaussian blur on GPU
gpu_blur = cv2.cuda.createGaussianFilter(cv2.CV_8UC1, cv2.CV_8UC1, (5, 5), 0)
gpu_blurred = gpu_blur.apply(gpu_gray)
# Apply Canny edge detection on GPU
gpu_canny = cv2.cuda.createCannyEdgeDetector(50, 150)
gpu_edges = gpu_canny.detect(gpu_blurred)
# Download edges from GPU
edges = gpu_edges.download()
else:
# CPU processing
gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
blur = cv2.GaussianBlur(gray, (5, 5), 0)
edges = cv2.Canny(blur, 50, 150)
# Define region of interest (ROI)
vertices = np.array([[
(int(width * 0.1), height),
(int(width * 0.45), int(height * 0.6)),
(int(width * 0.55), int(height * 0.6)),
(int(width * 0.9), height)
]], dtype=np.int32)
# Apply ROI mask
masked_edges = region_of_interest(edges, vertices)
# Apply Hough transform to detect lines
lines = cv2.HoughLinesP(
masked_edges,
rho=2,
theta=np.pi / 180,
threshold=50,
minLineLength=40,
maxLineGap=100
)
# Draw detected lanes on the original frame
if use_segmented:
result = draw_lines_segmented(frame.copy(), lines)
else:
result = draw_lines_basic(frame.copy(), lines)
return result
def calibrate_perspective(img):
"""
Apply perspective transform to get bird's eye view.
Converts trapezoidal ROI to rectangular view for easier lane detection.
"""
height, width = img.shape[:2]
# Define source points (trapezoid in original image)
src = np.float32([
[width * 0.45, height * 0.65], # Bottom left
[width * 0.55, height * 0.65], # Bottom right
[width * 0.9, height], # Top right
[width * 0.1, height] # Top left
])
# Define destination points (rectangle in bird's eye view)
dst = np.float32([
[width * 0.25, 0], # Top left
[width * 0.75, 0], # Top right
[width * 0.75, height], # Bottom right
[width * 0.25, height] # Bottom left
])
# Calculate perspective transform matrix
M = cv2.getPerspectiveTransform(src, dst)
# Calculate inverse perspective transform matrix
Minv = cv2.getPerspectiveTransform(dst, src)
# Apply perspective transform
warped = cv2.warpPerspective(img, M, (width, height), flags=cv2.INTER_LINEAR)
return warped, M, Minv
def color_and_gradient_threshold(img, use_enhanced=True):
"""
Apply color and gradient thresholding to isolate lane lines.
Enhanced version with better accuracy for various conditions.
Returns binary image with lane pixels set to 255.
"""
# Convert to HLS color space
hls = cv2.cvtColor(img, cv2.COLOR_BGR2HLS)
# Extract channels
h_channel = hls[:, :, 0]
l_channel = hls[:, :, 1]
s_channel = hls[:, :, 2]
# Enhanced thresholding for better lane detection
if use_enhanced:
# Adaptive thresholding for saturation channel
s_thresh = (90, 255) # Lower threshold for yellow lanes
s_binary = np.zeros_like(s_channel)
s_binary[(s_channel >= s_thresh[0]) & (s_channel <= s_thresh[1])] = 255
# Adaptive thresholding for lightness channel
l_thresh = (180, 255) # Lower threshold for white lanes
l_binary = np.zeros_like(l_channel)
l_binary[(l_channel >= l_thresh[0]) & (l_channel <= l_thresh[1])] = 255
# Apply CLAHE (Contrast Limited Adaptive Histogram Equalization) for better contrast
clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8, 8))
l_channel_enhanced = clahe.apply(l_channel.astype(np.uint8))
# Apply Sobel operator on enhanced channel
sobelx = cv2.Sobel(l_channel_enhanced, cv2.CV_64F, 1, 0, ksize=3)
abs_sobelx = np.absolute(sobelx)
scaled_sobel = np.uint8(255 * abs_sobelx / np.max(abs_sobelx))
# More sensitive gradient threshold
sobel_thresh = (15, 255) # Lower threshold for better edge detection
sobel_binary = np.zeros_like(scaled_sobel)
sobel_binary[(scaled_sobel >= sobel_thresh[0]) & (scaled_sobel <= sobel_thresh[1])] = 255
# Direction threshold to focus on vertical edges
sobely = cv2.Sobel(l_channel_enhanced, cv2.CV_64F, 0, 1, ksize=3)
abs_sobely = np.absolute(sobely)
scaled_sobely = np.uint8(255 * abs_sobely / np.max(abs_sobely))
# Calculate gradient direction
grad_dir = np.arctan2(abs_sobely, abs_sobelx)
dir_thresh = (0.7, 1.3) # Focus on near-vertical edges (lane lines)
dir_binary = np.zeros_like(scaled_sobel)
dir_binary[(grad_dir >= dir_thresh[0]) & (grad_dir <= dir_thresh[1])] = 255
# Combine all binary images with direction filter
combined_binary = np.zeros_like(s_binary)
combined_binary[((s_binary == 255) | (l_binary == 255) | (sobel_binary == 255)) & (dir_binary == 255)] = 255
# Apply morphological operations to reduce noise
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (3, 3))
combined_binary = cv2.morphologyEx(combined_binary, cv2.MORPH_CLOSE, kernel)
combined_binary = cv2.morphologyEx(combined_binary, cv2.MORPH_OPEN, kernel)
else:
# Basic thresholding
s_thresh = (100, 255)
s_binary = np.zeros_like(s_channel)
s_binary[(s_channel >= s_thresh[0]) & (s_channel <= s_thresh[1])] = 255
l_thresh = (200, 255)
l_binary = np.zeros_like(l_channel)
l_binary[(l_channel >= l_thresh[0]) & (l_channel <= l_thresh[1])] = 255
sobelx = cv2.Sobel(l_channel, cv2.CV_64F, 1, 0, ksize=3)
abs_sobelx = np.absolute(sobelx)
scaled_sobel = np.uint8(255 * abs_sobelx / np.max(abs_sobelx))
sobel_thresh = (20, 255)
sobel_binary = np.zeros_like(scaled_sobel)
sobel_binary[(scaled_sobel >= sobel_thresh[0]) & (scaled_sobel <= sobel_thresh[1])] = 255
combined_binary = np.zeros_like(s_binary)
combined_binary[(s_binary == 255) | (l_binary == 255) | (sobel_binary == 255)] = 255
return combined_binary
def fit_polynomial_lanes(binary_warped):
"""
Fit 2nd degree polynomials to lane lines using sliding window approach.
Returns left and right lane polynomial coefficients.
"""
# Take a histogram of the bottom half of the image
histogram = np.sum(binary_warped[binary_warped.shape[0]//2:, :], axis=0)
# Find the peak of the left and right halves of the histogram
midpoint = len(histogram) // 2
leftx_base = np.argmax(histogram[:midpoint])
rightx_base = np.argmax(histogram[midpoint:]) + midpoint
# Sliding window parameters
nwindows = 9
window_height = binary_warped.shape[0] // nwindows
margin = 100
minpix = 50
# Find nonzero pixels
nonzero = binary_warped.nonzero()
nonzeroy = np.array(nonzero[0])
nonzerox = np.array(nonzero[1])
# Current positions
leftx_current = leftx_base
rightx_current = rightx_base
# Lists to store lane pixel indices
left_lane_inds = []
right_lane_inds = []
# Step through windows
for window in range(nwindows):
# Window boundaries
win_y_low = binary_warped.shape[0] - (window + 1) * window_height
win_y_high = binary_warped.shape[0] - window * window_height
win_xleft_low = leftx_current - margin
win_xleft_high = leftx_current + margin
win_xright_low = rightx_current - margin
win_xright_high = rightx_current + margin
# Find pixels within windows
good_left_inds = ((nonzeroy >= win_y_low) & (nonzeroy < win_y_high) &
(nonzerox >= win_xleft_low) & (nonzerox < win_xleft_high)).nonzero()[0]
good_right_inds = ((nonzeroy >= win_y_low) & (nonzeroy < win_y_high) &
(nonzerox >= win_xright_low) & (nonzerox < win_xright_high)).nonzero()[0]
left_lane_inds.append(good_left_inds)
right_lane_inds.append(good_right_inds)
# Recenter windows
if len(good_left_inds) > minpix:
leftx_current = int(np.mean(nonzerox[good_left_inds]))
if len(good_right_inds) > minpix:
rightx_current = int(np.mean(nonzerox[good_right_inds]))
# Concatenate indices
left_lane_inds = np.concatenate(left_lane_inds)
right_lane_inds = np.concatenate(right_lane_inds)
# Extract pixel positions
leftx = nonzerox[left_lane_inds]
lefty = nonzeroy[left_lane_inds]
rightx = nonzerox[right_lane_inds]
righty = nonzeroy[right_lane_inds]
# Fit polynomial (2nd degree)
left_fit = None
right_fit = None
if len(leftx) > 0:
left_fit = np.polyfit(lefty, leftx, 2)
if len(rightx) > 0:
right_fit = np.polyfit(righty, rightx, 2)
return left_fit, right_fit
def draw_poly_lines(img, binary_warped, left_fit, right_fit, Minv):
"""
Draw polynomial lane lines on the original image using inverse perspective transform.
"""
if left_fit is None or right_fit is None:
return img
# Create an image to draw on
warp_zero = np.zeros_like(binary_warped).astype(np.uint8)
color_warp = np.dstack((warp_zero, warp_zero, warp_zero))
# Generate y values
ploty = np.linspace(0, binary_warped.shape[0] - 1, binary_warped.shape[0])
# Calculate x values using polynomial
left_fitx = left_fit[0] * ploty**2 + left_fit[1] * ploty + left_fit[2]
right_fitx = right_fit[0] * ploty**2 + right_fit[1] * ploty + right_fit[2]
# Ensure values are within image bounds
left_fitx = np.clip(left_fitx, 0, binary_warped.shape[1] - 1)
right_fitx = np.clip(right_fitx, 0, binary_warped.shape[1] - 1)
# Create points for the lane area
pts_left = np.array([np.transpose(np.vstack([left_fitx, ploty]))])
pts_right = np.array([np.flipud(np.transpose(np.vstack([right_fitx, ploty])))])
pts = np.hstack((pts_left, pts_right))
# Fill the lane area with green
cv2.fillPoly(color_warp, np.int_([pts]), (0, 255, 0))
# Draw the lane lines in red
cv2.polylines(color_warp, np.int32([pts_left]), isClosed=False, color=(0, 0, 255), thickness=15)
cv2.polylines(color_warp, np.int32([pts_right]), isClosed=False, color=(0, 0, 255), thickness=15)
# Apply inverse perspective transform to project back to original image
newwarp = cv2.warpPerspective(color_warp, Minv, (img.shape[1], img.shape[0]))
# Combine with original image
result = cv2.addWeighted(img, 0.8, newwarp, 0.5, 0)
return result
def process_frame_yolop(frame):
"""
YOLOP-inspired lane detection method.
Simulates multi-task learning approach with semantic segmentation.
Uses enhanced color-based segmentation with adaptive thresholding.
"""
height, width = frame.shape[:2]
# Convert to HLS for better color segmentation
hls = cv2.cvtColor(frame, cv2.COLOR_BGR2HLS)
h_channel = hls[:, :, 0]
l_channel = hls[:, :, 1]
s_channel = hls[:, :, 2]
# Multi-threshold approach for different lane colors
# White lanes - high lightness
white_mask = cv2.inRange(l_channel, 200, 255)
# Yellow lanes - specific hue range
yellow_mask = cv2.inRange(h_channel, 15, 35) & cv2.inRange(s_channel, 80, 255)
# Combine masks
color_mask = cv2.bitwise_or(white_mask, yellow_mask)
# Apply morphological operations
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (5, 5))
color_mask = cv2.morphologyEx(color_mask, cv2.MORPH_CLOSE, kernel)
color_mask = cv2.morphologyEx(color_mask, cv2.MORPH_OPEN, kernel)
# Apply ROI
vertices = np.array([[
(int(width * 0.1), height),
(int(width * 0.45), int(height * 0.6)),
(int(width * 0.55), int(height * 0.6)),
(int(width * 0.9), height)
]], dtype=np.int32)
color_mask = region_of_interest(color_mask, vertices)
# Find contours for lane segments
contours, _ = cv2.findContours(color_mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
# Create output image
result = frame.copy()
overlay = np.zeros_like(frame)
# Separate left and right lane contours
left_contours = []
right_contours = []
midpoint = width // 2
for contour in contours:
if cv2.contourArea(contour) > 100:
M = cv2.moments(contour)
if M["m00"] != 0:
cx = int(M["m10"] / M["m00"])
if cx < midpoint:
left_contours.append(contour)
else:
right_contours.append(contour)
# Draw lane regions
if len(left_contours) > 0 or len(right_contours) > 0:
# Fill lane area
if len(left_contours) > 0 and len(right_contours) > 0:
# Get bounding points
left_points = np.vstack(left_contours).squeeze()
right_points = np.vstack(right_contours).squeeze()
if len(left_points.shape) == 2 and len(right_points.shape) == 2:
# Sort by y coordinate
left_points = left_points[left_points[:, 1].argsort()]
right_points = right_points[right_points[:, 1].argsort()]
# Create polygon
poly_points = np.vstack([left_points, right_points[::-1]])
cv2.fillPoly(overlay, [poly_points], (0, 255, 0))
# Draw lane lines
for contour in left_contours:
cv2.drawContours(overlay, [contour], -1, (0, 0, 255), 5)
for contour in right_contours:
cv2.drawContours(overlay, [contour], -1, (0, 0, 255), 5)
# Blend with original
result = cv2.addWeighted(result, 0.8, overlay, 0.5, 0)
return result
def process_frame_ufld(frame):
"""
UFLD-inspired (Ultra Fast Lane Detection) method.
Uses row-wise classification approach with efficient feature extraction.
Focuses on speed and accuracy for real-time applications.
"""
height, width = frame.shape[:2]
# Convert to grayscale
gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
# Apply CLAHE for enhanced contrast
clahe = cv2.createCLAHE(clipLimit=3.0, tileGridSize=(8, 8))
enhanced = clahe.apply(gray)
# Apply bilateral filter to preserve edges while reducing noise
filtered = cv2.bilateralFilter(enhanced, 9, 75, 75)
# Adaptive thresholding
binary = cv2.adaptiveThreshold(
filtered, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C,
cv2.THRESH_BINARY, 11, 2
)
# Apply ROI
vertices = np.array([[
(int(width * 0.1), height),
(int(width * 0.45), int(height * 0.6)),
(int(width * 0.55), int(height * 0.6)),
(int(width * 0.9), height)
]], dtype=np.int32)
binary = region_of_interest(binary, vertices)
# Row-wise lane point detection
row_samples = 18 # Number of rows to sample
row_step = height // row_samples
left_lane_points = []
right_lane_points = []
midpoint = width // 2
for i in range(row_samples):
y = height - i * row_step - row_step // 2
if y < int(height * 0.6):
continue
row = binary[y, :]
# Find peaks in left and right halves
left_half = row[:midpoint]
right_half = row[midpoint:]
# Find lane positions
left_peaks = np.where(left_half > 200)[0]
right_peaks = np.where(right_half > 200)[0]
if len(left_peaks) > 0:
# Use the rightmost peak in left half
x = left_peaks[-1]
left_lane_points.append([x, y])
if len(right_peaks) > 0:
# Use the leftmost peak in right half
x = midpoint + right_peaks[0]
right_lane_points.append([x, y])
# Create result image
result = frame.copy()
overlay = np.zeros_like(frame)
# Fit curves to lane points
if len(left_lane_points) >= 3:
left_lane_points = np.array(left_lane_points)
left_fit = np.polyfit(left_lane_points[:, 1], left_lane_points[:, 0], 2)
# Generate smooth curve
ploty = np.linspace(int(height * 0.6), height, 100)
left_fitx = left_fit[0] * ploty**2 + left_fit[1] * ploty + left_fit[2]
left_fitx = np.clip(left_fitx, 0, width - 1)
left_curve = np.array([np.transpose(np.vstack([left_fitx, ploty]))], dtype=np.int32)
cv2.polylines(overlay, left_curve, False, (0, 0, 255), 8)
if len(right_lane_points) >= 3:
right_lane_points = np.array(right_lane_points)
right_fit = np.polyfit(right_lane_points[:, 1], right_lane_points[:, 0], 2)
# Generate smooth curve
ploty = np.linspace(int(height * 0.6), height, 100)
right_fitx = right_fit[0] * ploty**2 + right_fit[1] * ploty + right_fit[2]
right_fitx = np.clip(right_fitx, 0, width - 1)
right_curve = np.array([np.transpose(np.vstack([right_fitx, ploty]))], dtype=np.int32)
cv2.polylines(overlay, right_curve, False, (0, 0, 255), 8)
# Fill lane area
if len(left_lane_points) >= 3 and len(right_lane_points) >= 3:
ploty = np.linspace(int(height * 0.6), height, 100)
left_fitx = left_fit[0] * ploty**2 + left_fit[1] * ploty + left_fit[2]
right_fitx = right_fit[0] * ploty**2 + right_fit[1] * ploty + right_fit[2]
left_fitx = np.clip(left_fitx, 0, width - 1)
right_fitx = np.clip(right_fitx, 0, width - 1)
pts_left = np.array([np.transpose(np.vstack([left_fitx, ploty]))])
pts_right = np.array([np.flipud(np.transpose(np.vstack([right_fitx, ploty])))])
pts = np.hstack((pts_left, pts_right))
cv2.fillPoly(overlay, np.int32([pts]), (0, 255, 0))
# Blend
result = cv2.addWeighted(result, 0.8, overlay, 0.5, 0)
return result
def process_frame_scnn(frame):
"""
SCNN-inspired (Spatial CNN) method.
Uses spatial message passing for lane detection.
Implements slice-by-slice convolutions in four directions.
"""
height, width = frame.shape[:2]
# Preprocessing
hls = cv2.cvtColor(frame, cv2.COLOR_BGR2HLS)
l_channel = hls[:, :, 1]
s_channel = hls[:, :, 2]
# Enhanced preprocessing with CLAHE
clahe = cv2.createCLAHE(clipLimit=2.5, tileGridSize=(8, 8))
l_enhanced = clahe.apply(l_channel)
# Multi-scale edge detection
sobel_x = cv2.Sobel(l_enhanced, cv2.CV_64F, 1, 0, ksize=5)
sobel_y = cv2.Sobel(l_enhanced, cv2.CV_64F, 0, 1, ksize=5)
# Gradient magnitude and direction
magnitude = np.sqrt(sobel_x**2 + sobel_y**2)
magnitude = np.uint8(255 * magnitude / np.max(magnitude))
direction = np.arctan2(sobel_y, sobel_x)
# Focus on near-vertical edges (lane lines)
vertical_mask = np.zeros_like(magnitude)
vertical_mask[(np.abs(direction) > 0.6) & (np.abs(direction) < 1.5)] = 255
# Combine with color thresholding
s_binary = cv2.inRange(s_channel, 90, 255)
l_binary = cv2.inRange(l_enhanced, 180, 255)
combined = cv2.bitwise_or(s_binary, l_binary)
combined = cv2.bitwise_and(combined, magnitude)
combined = cv2.bitwise_and(combined, vertical_mask)
# Simulate spatial message passing with directional filtering
# Horizontal message passing (left-to-right and right-to-left)
kernel_h = np.ones((1, 15), np.uint8)
horizontal_pass = cv2.morphologyEx(combined, cv2.MORPH_CLOSE, kernel_h)
# Vertical message passing (top-to-bottom and bottom-to-top)
kernel_v = np.ones((15, 1), np.uint8)
spatial_features = cv2.morphologyEx(horizontal_pass, cv2.MORPH_CLOSE, kernel_v)
# Apply ROI
vertices = np.array([[
(int(width * 0.1), height),
(int(width * 0.45), int(height * 0.6)),
(int(width * 0.55), int(height * 0.6)),
(int(width * 0.9), height)
]], dtype=np.int32)
spatial_features = region_of_interest(spatial_features, vertices)
# Lane fitting with sliding window
histogram = np.sum(spatial_features[spatial_features.shape[0]//2:, :], axis=0)
midpoint = len(histogram) // 2
leftx_base = np.argmax(histogram[:midpoint])
rightx_base = np.argmax(histogram[midpoint:]) + midpoint
# Sliding window parameters
nwindows = 12
window_height = spatial_features.shape[0] // nwindows
margin = 80
minpix = 40
nonzero = spatial_features.nonzero()
nonzeroy = np.array(nonzero[0])
nonzerox = np.array(nonzero[1])
leftx_current = leftx_base
rightx_current = rightx_base
left_lane_inds = []
right_lane_inds = []
for window in range(nwindows):
win_y_low = spatial_features.shape[0] - (window + 1) * window_height
win_y_high = spatial_features.shape[0] - window * window_height
win_xleft_low = leftx_current - margin
win_xleft_high = leftx_current + margin
win_xright_low = rightx_current - margin
win_xright_high = rightx_current + margin
good_left_inds = ((nonzeroy >= win_y_low) & (nonzeroy < win_y_high) &
(nonzerox >= win_xleft_low) & (nonzerox < win_xleft_high)).nonzero()[0]
good_right_inds = ((nonzeroy >= win_y_low) & (nonzeroy < win_y_high) &
(nonzerox >= win_xright_low) & (nonzerox < win_xright_high)).nonzero()[0]
left_lane_inds.append(good_left_inds)
right_lane_inds.append(good_right_inds)
if len(good_left_inds) > minpix:
leftx_current = int(np.mean(nonzerox[good_left_inds]))
if len(good_right_inds) > minpix:
rightx_current = int(np.mean(nonzerox[good_right_inds]))
left_lane_inds = np.concatenate(left_lane_inds)
right_lane_inds = np.concatenate(right_lane_inds)
leftx = nonzerox[left_lane_inds]
lefty = nonzeroy[left_lane_inds]
rightx = nonzerox[right_lane_inds]
righty = nonzeroy[right_lane_inds]
result = frame.copy()
overlay = np.zeros_like(frame)
if len(leftx) > 0 and len(rightx) > 0:
left_fit = np.polyfit(lefty, leftx, 2)
right_fit = np.polyfit(righty, rightx, 2)
ploty = np.linspace(0, spatial_features.shape[0] - 1, spatial_features.shape[0])
left_fitx = left_fit[0] * ploty**2 + left_fit[1] * ploty + left_fit[2]
right_fitx = right_fit[0] * ploty**2 + right_fit[1] * ploty + right_fit[2]
left_fitx = np.clip(left_fitx, 0, width - 1)
right_fitx = np.clip(right_fitx, 0, width - 1)
# Draw lane area
pts_left = np.array([np.transpose(np.vstack([left_fitx, ploty]))])
pts_right = np.array([np.flipud(np.transpose(np.vstack([right_fitx, ploty])))])
pts = np.hstack((pts_left, pts_right))
cv2.fillPoly(overlay, np.int32([pts]), (0, 255, 0))
# Draw lane lines
cv2.polylines(overlay, np.int32([pts_left]), False, (0, 0, 255), 12)
cv2.polylines(overlay, np.int32([pts_right]), False, (0, 0, 255), 12)
result = cv2.addWeighted(result, 0.8, overlay, 0.5, 0)
return result
def process_frame(frame, method="advanced", use_enhanced=True, use_segmented=False):
"""
Process a single frame for lane detection.
method: "basic", "basic_segmented", "advanced", "yolop", "ufld", "scnn"
use_enhanced: Use enhanced thresholding for better accuracy (advanced method only)
use_segmented: Use segmented lines for curve representation (basic method only)
"""
if method == "basic" or method == "basic_standard":
return process_frame_basic(frame, use_segmented=False)
elif method == "basic_segmented":
return process_frame_basic(frame, use_segmented=True)
elif method == "advanced":
return process_frame_advanced(frame, use_enhanced)
elif method == "yolop":
return process_frame_yolop(frame)
elif method == "ufld":
return process_frame_ufld(frame)
elif method == "scnn":
return process_frame_scnn(frame)
else:
raise ValueError(f"Unknown method: {method}. Use 'basic', 'basic_segmented', 'advanced', 'yolop', 'ufld', or 'scnn'")
def process_frame_advanced(frame, use_enhanced=True):
"""
Process a single frame for lane detection using advanced pipeline.
1. Perspective transform to bird's eye view
2. Enhanced color and gradient thresholding
3. Polynomial fitting with sliding windows
4. Draw lanes with inverse perspective transform
"""
# Step 1: Apply perspective transform to get bird's eye view
warped, M, Minv = calibrate_perspective(frame)
# Step 2: Apply enhanced color and gradient thresholding
binary_warped = color_and_gradient_threshold(warped, use_enhanced)
# Step 3: Fit polynomial lanes using sliding window approach
left_fit, right_fit = fit_polynomial_lanes(binary_warped)
# Step 4: Draw polynomial lines on original image
result = draw_poly_lines(frame, binary_warped, left_fit, right_fit, Minv)
return result
def process_video(input_path, output_path, method="advanced", use_enhanced=True, use_segmented=False, progress_callback=None):
"""
Process the video and create side-by-side comparison.
method: "basic", "basic_segmented", "advanced", "yolop", "ufld", "scnn"
use_enhanced: Use enhanced thresholding for better accuracy (advanced method only)
use_segmented: Use segmented lines for curve representation (basic method only)
progress_callback: Optional callback function to report progress (value between 0 and 1)
Returns True if successful, False otherwise.
"""
# Open the video
cap = cv2.VideoCapture(input_path)
if not cap.isOpened():
return False
# Get video properties
fps = int(cap.get(cv2.CAP_PROP_FPS))
width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
total_frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
# Video writer for output (side-by-side, so width is doubled)
# Try different codecs for better browser compatibility
codecs_to_try = [
('H264', cv2.VideoWriter_fourcc(*'H264')), # H.264 - most compatible
('h264', cv2.VideoWriter_fourcc(*'h264')), # Alternative H.264
('mp4v', cv2.VideoWriter_fourcc(*'mp4v')), # MPEG-4 Part 2
('XVID', cv2.VideoWriter_fourcc(*'XVID')), # XVID
]
out = None
selected_codec = None
for codec_name, fourcc in codecs_to_try:
out = cv2.VideoWriter(output_path, fourcc, fps, (width * 2, height))
if out.isOpened():
selected_codec = codec_name
print(f"✓ Using codec: {codec_name}")
break
else:
print(f"⚠ Codec {codec_name} not available, trying next...")
if out is None or not out.isOpened():
print("✗ Error: No suitable video codec found!")
return False
frame_count = 0
print(f"Processing {total_frames} frames using {method} method...")
if method == "advanced" and use_enhanced:
print("Enhanced thresholding enabled for better accuracy")
if method == "basic" and use_segmented:
print("Segmented line mode enabled for better curve representation")
# Process each frame
while True:
ret, frame = cap.read()
if not ret:
break
# Process frame for lane detection
processed_frame = process_frame(frame, method, use_enhanced, use_segmented)
# Create side-by-side comparison
# Original on left, processed on right
combined = np.hstack((frame, processed_frame))
# Write the combined frame
out.write(combined)
frame_count += 1
# Progress indicator
if progress_callback and frame_count % 10 == 0:
progress = frame_count / total_frames if total_frames > 0 else 0
progress_callback(progress, f"Processing frame {frame_count}/{total_frames}")
elif frame_count % 30 == 0:
progress = (frame_count / total_frames) * 100 if total_frames > 0 else 0
print(f"Progress: {frame_count}/{total_frames} frames ({progress:.1f}%)")
# Release resources
cap.release()
out.release()
if progress_callback:
progress_callback(1.0, "Completed!")
print(f"✓ Completed! Processed {frame_count} frames using {method} method.")
return frame_count > 0
|