def showHOGfeature(car, notcar):
features, car_dst = get_hog_features(car[:,:,2], 9, 8, 8, vis=True, feature_vec=True)
features, noncar_dst = get_hog_features(notcar[:,:,2], 9, 8, 8, vis=True, feature_vec=True)
f, ((ax1, ax2), (ax3, ax4)) = plt.subplots(2, 2, figsize=(7,7))
f.subplots_adjust(hspace = .4, wspace=.2)
ax1.imshow(car)
ax1.set_title('Car Image', fontsize=16)
ax2.imshow(car_dst, cmap='gray')
ax2.set_title('Car HOG', fontsize=16)
ax3.imshow(notcar)
ax3.set_title('Non-Car Image', fontsize=16)
ax4.imshow(noncar_dst, cmap='gray')
ax4.set_title('Non-Car HOG', fontsize=16)
def sample_hog(detect: VehicleDetector, cars, notcars):
# Generate a random index to look at a car image
ind = np.random.randint(0, len(cars))
# Read in the image
image = mpimg.imread(cars[ind])
image = image.astype(np.float32)
image = image[:, :, 0]
# Call our function with vis=True to see an image output
features, hog_image = functions.get_hog_features(image, orient=detect.orient,
pix_per_cell=detect.pix_per_cell,
cell_per_block=detect.cell_per_block,
hog_vis=detect.hog_vis, hog_feature_vec=detect.hog_feature_vec)
# Plot the examples
fig = plt.figure()
plt.subplot(121)
plt.imshow(image, cmap='gray')
plt.title('Example Car Image ')
plt.subplot(122)
plt.imshow(hog_image, cmap='hot')
plt.title('HOG Visualization')
fig.savefig('./output_images/example_hog.jpg')
plt.show()
def find_cars(img, x_start_stop, y_start_stop, scale,color_space ,svc, X_scaler, orient,
pix_per_cell, cell_per_block, spatial_size, hist_bins, show_all_rectangles):
img = img.astype(np.float32)/255
box_list = []
if x_start_stop[0] == None:
x_start_stop[0] = 0
if x_start_stop[1] == None:
x_start_stop[1] = img.shape[1]
if y_start_stop[0] == None:
y_start_stop[0] = 0
if y_start_stop[1] == None:
y_start_stop[1] = img.shape[0]
img_tosearch = img[y_start_stop[0]:y_start_stop[1],x_start_stop[0]:x_start_stop[1],:]
# ctrans_tosearch = cv2.cvtColor(img_tosearch, cv2.COLOR_RGB2YCrCb)
if color_space != 'RGB':
if color_space == 'HSV':
ctrans_tosearch = cv2.cvtColor(img_tosearch, cv2.COLOR_RGB2HSV)
elif color_space == 'LUV':
ctrans_tosearch = cv2.cvtColor(img_tosearch, cv2.COLOR_RGB2LUV)
elif color_space == 'HLS':
ctrans_tosearch = cv2.cvtColor(img_tosearch, cv2.COLOR_RGB2HLS)
elif color_space == 'YUV':
ctrans_tosearch = cv2.cvtColor(img_tosearch, cv2.COLOR_RGB2YUV)
elif color_space == 'YCrCb':
ctrans_tosearch = cv2.cvtColor(img_tosearch, cv2.COLOR_RGB2YCrCb)
#ctrans_tosearch = ctrans_tosearch.astype(np.float32)/255
else: ctrans_tosearch = np.copy(img_tosearch)
#print(ctrans_tosearch.shape)
if scale != 1:
imshape = ctrans_tosearch.shape
ctrans_tosearch = cv2.resize(ctrans_tosearch, (np.int(imshape[1]/scale), np.int(imshape[0]/scale)))
ch1 = ctrans_tosearch[:,:,0]
ch2 = ctrans_tosearch[:,:,1]
ch3 = ctrans_tosearch[:,:,2]
# Define blocks and steps as above
nxblocks = (ch1.shape[1] // pix_per_cell) - cell_per_block + 1
nyblocks = (ch1.shape[0] // pix_per_cell) - cell_per_block + 1
#nfeat_per_block = orient*cell_per_block**2
# 64 was the orginal sampling rate, with 8 cells and 8 pix per cell
window = 64
nblocks_per_window = (window // pix_per_cell) - cell_per_block + 1
cells_per_step = 2 # Instead of overlap, define how many cells to step
nxsteps = (nxblocks - nblocks_per_window) // cells_per_step
nysteps = (nyblocks - nblocks_per_window) // cells_per_step
# Compute individual channel HOG features for the entire image
if hog_channel == 'ALL':
hog1 = get_hog_features(ch1, orient, pix_per_cell, cell_per_block, feature_vec=False)
hog2 = get_hog_features(ch2, orient, pix_per_cell, cell_per_block, feature_vec=False)
hog3 = get_hog_features(ch3, orient, pix_per_cell, cell_per_block, feature_vec=False)
else:
hog1 = get_hog_features(ch1, orient, pix_per_cell, cell_per_block, feature_vec=False)
for xb in range(nxsteps):
for yb in range(nysteps):
ypos = yb*cells_per_step
xpos = xb*cells_per_step
# Extract HOG for this patch
if hog_channel == 'ALL':
hog_feat1 = hog1[ypos:ypos+nblocks_per_window, xpos:xpos+nblocks_per_window].ravel()
hog_feat2 = hog2[ypos:ypos+nblocks_per_window, xpos:xpos+nblocks_per_window].ravel()
hog_feat3 = hog3[ypos:ypos+nblocks_per_window, xpos:xpos+nblocks_per_window].ravel()
hog_features = np.hstack((hog_feat1, hog_feat2, hog_feat3))
else:
hog_feat1 = hog1[ypos:ypos+nblocks_per_window, xpos:xpos+nblocks_per_window].ravel()
hog_features = hog_feat1
xleft = xpos*pix_per_cell
ytop = ypos*pix_per_cell
# Extract the image patch
subimg = cv2.resize(ctrans_tosearch[ytop:ytop+window, xleft:xleft+window], (64,64))
# Get color features
spatial_features = bin_spatial(subimg, size=spatial_size)
hist_features = color_hist(subimg, nbins=hist_bins)
features = np.hstack((spatial_features, hist_features, hog_features))
#print(np.shape(features))
# Scale features and make a prediction
features[np.isnan(features) == True] = 0
test_features = X_scaler.transform(np.array(features).reshape(1, -1))
#test_features = X_scaler.transform(np.hstack((hog_features)).reshape(1, -1))
test_prediction = svc.predict(test_features)
if test_prediction == 1 or show_all_rectangles:
xbox_left = np.int(xleft*scale)
ytop_draw = np.int(ytop*scale)
win_draw = np.int(window*scale)
box_list.append([(xbox_left+x_start_stop[0], ytop_draw+y_start_stop[0]),(xbox_left+win_draw+x_start_stop[0],ytop_draw+win_draw+y_start_stop[0])])
return box_list
# axh[3].set_title('Car V HOG', fontsize=10)
#
#
# axh[4].imshow(notcrgb)
# axh[4].set_title('Not-Car Image', fontsize=10)
# axh[5].imshow(notcar_hog_y, cmap='gray')
# axh[5].set_title('Not-Car Y HOG', fontsize=10)
# axh[6].imshow(notcar_hog_u, cmap='gray')
# axh[6].set_title('Not-Car U HOG', fontsize=10)
# axh[7].imshow(notcar_hog_v, cmap='gray')
# axh[7].set_title('Not-Car V HOG', fontsize=10)
#ycrcb sapce
fcar_y, car_hog_y = get_hog_features(cycrcb[:, :, 0],
orient,
pix_per_cell,
cells_per_block,
vis=True,
feature_vec=True)
fnotcar_y, notcar_hog_y = get_hog_features(notcycrcb[:, :, 0],
orient,
pix_per_cell,
cells_per_block,
vis=True,
feature_vec=True)
fcar_u, car_hog_u = get_hog_features(cycrcb[:, :, 0],
orient,
pix_per_cell,
8,
vis=True,
feature_vec=True)
ch1 = ctrans_tosearch[:, :, 0]
ch2 = ctrans_tosearch[:, :, 1]
ch3 = ctrans_tosearch[:, :, 2]
nxblocks = (ch1.shape[1] // pix_per_cell) - 1
nyblocks = (ch1.shape[0] // pix_per_cell) - 1
nfeat_per_block = orient * cell_per_block**2
window = 64
nblocks_per_window = (window // pix_per_cell) - 1
cells_per_step = 2
nxsteps = (nxblocks - nblocks_per_window) // cells_per_step
nysteps = (nyblocks - nblocks_per_window) // cells_per_step
hog1 = functions.get_hog_features(ch1,
orient,
pix_per_cell,
cell_per_block,
feature_vec=False)
hog2 = functions.get_hog_features(ch2,
orient,
pix_per_cell,
cell_per_block,
feature_vec=False)
hog3 = functions.get_hog_features(ch3,
orient,
pix_per_cell,
cell_per_block,
feature_vec=False)
for xb in range(nxsteps):
for yb in range(nysteps):
d = fig3.add_subplot(3, 2, 4)
plt.bar(bin_centers, notcar_hist2[0])
d.set_title("not car-YCrCB ch2 histogram")
e = fig3.add_subplot(3, 2, 5)
plt.bar(bin_centers, car_hist3[0])
e.set_title("car-YCrCB ch3 histogram")
f = fig3.add_subplot(3, 2, 6)
plt.bar(bin_centers, notcar_hist3[0])
f.set_title("not car-YCrCB ch3 histogram")
fig3.savefig('./examples/car_notcar_histogram.jpg')
fig4 = plt.figure(figsize=(10, 12))
a = fig4.add_subplot(3, 2, 1)
hog, img = get_hog_features(car_sample_YCrCb[:, :, 0],
orient=9,
pix_per_cell=8,
cell_per_block=1,
vis=True,
feature_vec=False)
plt.imshow(img, cmap='gray')
a.set_title("car-YCrCB ch1 hog")
b = fig4.add_subplot(3, 2, 2)
hog, img = get_hog_features(not_car_sample_YCrCb[:, :, 0],
orient=9,
pix_per_cell=8,
cell_per_block=1,
vis=True,
feature_vec=False)
plt.imshow(img, cmap='gray')
b.set_title("not car-YCrCB ch1 hog")
c = fig4.add_subplot(3, 2, 3)
hog, img = get_hog_features(car_sample_YCrCb[:, :, 1],