def search_windows(img,
windows,
clf,
scaler,
color_space='RGB',
patch_size=(64, 64),
feature_extractor_params=None):
if color_space != 'RGB':
img = convert_color(img, color_space)
# 1) Create an empty list to receive positive detection windows
on_windows = []
# 2) Iterate over all windows in the list
for window in windows:
# 3) Extract the test window from original image
test_img = cv2.resize(
img[window[0][1]:window[1][1], window[0][0]:window[1][0]],
patch_size)
# 4) Extract features for that window using single_img_features()
features = extract_features(test_img, **feature_extractor_params)
# 5) Scale extracted features to be fed to classifier
test_features = scaler.transform(np.array(features).reshape(1, -1))
# 6) Predict using your classifier
prediction = clf.predict(test_features)
# 7) If positive (prediction == 1) then save the window
if prediction == 1:
on_windows.append(window)
# 8) Return windows for positive detections
return on_windows
def extract_features_imgs(imgs, cspace='RGB', **params):
"""Extract features from a list of images
Useful for extracting features from training and test images
:param imgs:
:param bin_params: parameters for spatial binning
:param color_params: parameters for color histogram binning
:param hog_params: parameters for hog feature extractor
:return:
"""
# Create a list to append feature vectors to
features = []
# Iterate through the list of images
# Read in each one by one
# apply color conversion if other than 'RGB'
# Apply bin_spatial() to get spatial color features
# Apply color_hist() to get color histogram features
# Get HOG features
# Append the new feature vector to the features list
# Return list of feature vectors
for image in imgs:
if type(image) is str:
image = imread(image)
if cspace is not 'RGB':
image = convert_color(image, cspace)
features.append(extract_features(image, **params))
return features
def find_cars(img, ystart, ystop, scale, clf, X_scaler, cspace, spatial_size, hist_bins,
orient, pix_per_cell, cell_per_block, hog_channel):
'''Detect vehicles and return containing boxes'''
img = img.astype(np.float32)/255
img_tosearch = img[ystart:ystop, :, :]
ctrans_tosearch = convert_color(img_tosearch, cspace=cspace)
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
# 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 + 1
nysteps = (nyblocks - nblocks_per_window) // cells_per_step + 1
# 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:
channels = [ch1, ch2, ch3]
hog = get_hog_features(channels[hog_channel], orient, pix_per_cell, cell_per_block, feature_vec=False)
boxes = []
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_features = hog[ypos:ypos+nblocks_per_window, xpos:xpos+nblocks_per_window].ravel()
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)
# Scale features and make a prediction
test_features = X_scaler.transform(np.hstack((spatial_features, hist_features, hog_features)).reshape(1, -1))
test_prediction = clf.predict(test_features)
if test_prediction == 1:
xbox_left = np.int(xleft*scale)
ytop_draw = np.int(ytop*scale)
win_draw = np.int(window*scale)
boxes.append(((xbox_left, ytop_draw+ystart), (xbox_left+win_draw, ytop_draw+win_draw+ystart)))
return boxes
def extract_features_from_files(img_files, **kwargs):
"""
extract features from a list of images
:param img_files:
:param kwargs:
:return:
"""
color_conversion = kwargs.get("color_conversion", 'RGB2YCrCb')
orient = kwargs.get("orient", 9)
pix_per_cell = kwargs.get("pix_per_cell", 8)
cell_per_block = kwargs.get("cell_per_block", 2)
hog_channel = kwargs.get("hog_channel", 0)
spatial_size = kwargs.get("spatial_size", (32, 32))
hist_bins = kwargs.get("hist_bins", 32)
# Create a list to append feature vectors to
features = []
# Iterate through the list of images
for file in img_files:
image = mpimg.imread(file)
# apply color conversion if other than 'RGB'
feature_image = convert_color(image, conv=color_conversion)
# Call get_hog_features() with vis=False, feature_vec=True
if hog_channel == 'ALL':
hog_features = []
for channel in range(feature_image.shape[2]):
hog_features.append(
get_hog_features(feature_image[:, :, channel],
orient,
pix_per_cell,
cell_per_block,
visualize=False,
feature_vec=True))
hog_features = np.ravel(hog_features)
else:
hog_features = get_hog_features(feature_image[:, :, hog_channel],
orient,
pix_per_cell,
cell_per_block,
visualize=False,
feature_vec=True)
# Extract the image patch
sub_img = cv2.resize(feature_image, (64, 64))
# Get color features
spatial_features = bin_spatial(sub_img, size=spatial_size)
hist_features = color_hist(sub_img, nbins=hist_bins)
image_features = np.hstack(
(hog_features, spatial_features, hist_features))
# Append the new feature vector to the features list
features.append(image_features)
return features
def extract_feature(image,
color_space='RGB',
spatial_size=(32, 32),
hist_bins=32,
orient=9,
pix_per_cell=8,
cell_per_block=2,
hog_channel=0,
spatial_feat=True,
hist_feat=True,
hog_feat=True):
"""
Extract features of an image
"""
# Create a list to contain different feature for later concatenation
sub_features = []
feature_image = tu.convert_color(image, color_space)
# Extract spatial binning features
if spatial_feat:
spatial_features = fx.bin_spatial(feature_image, size=spatial_size)
sub_features.append(spatial_features)
# Extract color histogram features
if hist_feat:
hist_features = fx.color_hist(feature_image, nbins=hist_bins)
sub_features.append(hist_features)
# Extract HOG features
if hog_feat:
if hog_channel == 'ALL':
hog_features = []
for channel in range(feature_image.shape[2]):
hog_features.append(
fx.get_hog_features(feature_image[:, :, channel],
orient,
pix_per_cell,
cell_per_block,
vis=False,
feature_vec=True))
hog_features = np.ravel(hog_features)
else:
hog_features = fx.get_hog_features(feature_image[:, :,
hog_channel],
orient,
pix_per_cell,
cell_per_block,
vis=False,
feature_vec=True)
sub_features.append(hog_features)
return np.concatenate(sub_features)
def format_draw(card: Dict, who: int, order: int) -> Optional[str]:
""" Json message server sends when card is dealt """
def convert_rank(rank):
if rank is None:
return -1
elif rank > -1:
return rank + 1
else:
return rank
suit = utils.convert_color(card['color'])
rank = convert_rank(card['rank'])
who = str(who)
order = str(order)
return '{{"type":"draw","who":{0},"rank":{1},"suit":{2},"order":{3}}}'.format(
who, rank, suit, order)
def transform(self, X):
print(X.shape)
params = self.__construct_params(**self.get_params(deep=False))
feature_list = []
for idx in range(X.shape[0]):
#if idx % 1000 == 0:
# print('iter', idx)
im = np.squeeze(X[idx])
if params['cspace'] != 'RGB':
im = convert_color(im, params['cspace'])
feature_list.append(self.extract_feature(im, **params))
print('transform done...scaling...')
feature_vec = np.vstack(feature_list).astype(np.float64)
# Fit a per-column scaler
X_scaler = StandardScaler().fit(feature_vec)
# Apply the scaler to X
scaled_X = X_scaler.transform(feature_vec)
print('scaling done ', scaled_X.shape)
return scaled_X
def extract_image_features(self, img):
"""
Extract features from single image
"""
features = []
cvt_img = convert_color(img, self.P.color_space)
spatial_features = get_spatial_features(cvt_img, size=self.P.spatial_size)
features.append(spatial_features)
color_features = get_color_features(cvt_img, size=self.P.window_size,
nbins=self.P.color_nbins)
features.append(color_features)
if self.P.window_size != (cvt_img.shape[0], cvt_img.shape[1]):
cvt_img = cv2.resize(cvt_img, self.P.window_size)
hog_features = get_hog_features(cvt_img, orient=self.P.orient,
pix_per_cell=self.P.pix_per_cell,
cell_per_block=self.P.cell_per_block)
features.append(hog_features)
return np.concatenate(features)
def fill(self, color, rect = None):
color = utils.convert_color(color)
self._pixels.fill(color, rect)
def search_windows_v2(img,
windows,
clf,
scaler,
color_space='RGB',
spatial_size=(32, 32),
hist_bins=32,
hist_range=(0, 256),
orient=9,
pix_per_cell=8,
cell_per_block=2,
hog_channel=0,
spatial_feat=True,
hist_feat=True,
hog_feat=True,
y_start_stop=[None, None],
scale=0):
draw_img = np.copy(img)
img = img.astype(np.float32) / 255
img_tosearch = img[y_start_stop[0]:y_start_stop[1], :, :]
ctrans_tosearch = utils.convert_color(img_tosearch, conv='RGB2YCrCb')
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
hog1 = utils.get_hog_features(ch1,
orient,
pix_per_cell,
cell_per_block,
feature_vec=False)
hog2 = utils.get_hog_features(ch2,
orient,
pix_per_cell,
cell_per_block,
feature_vec=False)
hog3 = utils.get_hog_features(ch3,
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
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))
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
features = utils.single_img_features(subimg,
color_space=color_space,
spatial_size=spatial_size,
hist_bins=hist_bins,
orient=orient,
pix_per_cell=pix_per_cell,
cell_per_block=cell_per_block,
hog_channel=hog_channel,
spatial_feat=spatial_feat,
hist_feat=hist_feat,
hog_feat=hog_feat)
X = scaler.transform(np.array(features).reshape(1, -1))
# 6) Predict using your classifier
prediction = clf.predict(X)
# Scale features and make a prediction
# test_features = X_scaler.transform(X)
# test_features = X_scaler.transform(np.hstack((shape_feat, hist_feat)).reshape(1, -1))
# test_prediction = clf.predict(X)
if prediction == 1:
xbox_left = np.int(xleft * scale)
ytop_draw = np.int(ytop * scale)
win_draw = np.int(window * scale)
cv2.rectangle(draw_img,
(xbox_left, ytop_draw + y_start_stop[0]),
(xbox_left + win_draw,
ytop_draw + win_draw + y_start_stop[0]),
(0, 0, 255), 6)
return draw_img
def find_cars(img, ystart, ystop, scale, svc, X_scaler, orient, pix_per_cell, cell_per_block, spatial_size, hist_bins):
draw_img = np.copy(img)
img = img.astype(np.float32) / 255
img_tosearch = img[ystart:ystop, :, :]
ctrans_tosearch = utils.convert_color(img_tosearch, conv='RGB2YCrCb')
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
hog1 = utils.get_hog_features(ch1, orient, pix_per_cell, cell_per_block, feature_vec=False)
hog2 = utils.get_hog_features(ch2, orient, pix_per_cell, cell_per_block, feature_vec=False)
hog3 = utils.get_hog_features(ch3, 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
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))
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 = utils.bin_spatial(subimg, size=spatial_size)
hist_features = utils.color_hist(subimg, nbins=hist_bins)
print(hist_features.shape)
print(spatial_features.shape)
print(hog_features.shape)
X = np.hstack((spatial_features, hist_features, hog_features)).reshape(-1, 1)
print(X.shape)
# Scale features and make a prediction
# test_features = X_scaler.transform(X)
# test_features = X_scaler.transform(np.hstack((shape_feat, hist_feat)).reshape(1, -1))
test_prediction = svc.predict(X)
if test_prediction == 1:
xbox_left = np.int(xleft * scale)
ytop_draw = np.int(ytop * scale)
win_draw = np.int(window * scale)
cv2.rectangle(draw_img, (xbox_left, ytop_draw + ystart),
(xbox_left + win_draw, ytop_draw + win_draw + ystart), (0, 0, 255), 6)
return draw_img
[416, 480, 1.0],
[400, 480, 1.25],
[424, 504, 1.25],
[400, 496, 1.5],
[432, 528, 1.5],
[400, 512, 1.75],
[432, 544, 1.75],
[400, 528, 2.0],
[432, 560, 2.0],
[400, 596, 3.5],
[464, 660, 3.5]]
for i, img in enumerate([car_example, notcar_example]):
plt.imsave(r"./output_images/" + str(i) + "_1_spatial.jpg",
cv2.resize(cv2.resize(img, spatial_size), (128, 128)))
img2 = convert_color(img, cspace=cspace)
ch1 = img2[:, :, 0]
ch2 = img2[:, :, 1]
ch3 = img2[:, :, 2]
plt.imsave(r"./output_images/" + str(i) + "_2_ch1.jpg", ch1, cmap='gray')
plt.imsave(r"./output_images/" + str(i) + "_2_ch2.jpg", ch2, cmap='gray')
plt.imsave(r"./output_images/" + str(i) + "_2_ch3.jpg", ch3, cmap='gray')
hog_feats, hog_im = get_hog_features(img, orient, pix_per_cell, cell_per_block,
vis=True, feature_vec=True)
plt.imsave(r"./output_images/" + str(i) + "_3_hog.jpg", hog_im, cmap='gray')
for fname in images:
print('processing ', fname, '...')
img = mpimg.imread(fname)
boxes = find_cars_multiscale(img, multiscale, clf, X_scaler, cspace, spatial_size,
# idx = list(r['class_ids']).index(2)
images = []
idx_s = []
while True:
idx = input()
if idx == 'q':
break
idx_s.append(int(idx))
idx = int(idx)
res = crop_by_id(img, masks, idx)
# display(res)
final_image = img.copy()
final_img = convert_color(img, res, rgb)
images.append(final_img)
print('enter q to break, else enter index \n')
'''
final_image = img.copy()
final_image[:,:,0] = np.where(masks[:,:,4], np.zeros((512,512)), final_image[:,:,0])
final_image[:,:,1] = np.where(masks[:,:,4], np.zeros((512,512)), final_image[:,:,1])
final_image[:,:,2] = np.where(masks[:,:,4], np.zeros((512,512)), final_image[:,:,2])
display(final_image)
'''
l = len(images)
if l == 1:
for idx, i in enumerate(images):
def draw_line(self, p1, p2, color, thickness = 1):
color = utils.convert_color(color)
pygame.draw.line(p1, p2, color, thickness)
label_descriptions = json.load(f)
vid = cv2.VideoCapture(0)
vid.set(cv2.CAP_PROP_FRAME_WIDTH, 1200)
vid.set(cv2.CAP_PROP_FRAME_HEIGHT, 800)
while (True):
ret, frame = vid.read()
try:
img, r = masker_np(frame)
masks, rois = convert_to_mask(img, r)
idx = list(r['class_ids']).index(2)
res = crop_by_id(img, masks, idx)
final_img = convert_color(img, res, [244, 196, 48])
img[:, :, 0] = np.where(masks[:, :, idx], np.zeros((512, 512)),
img[:, :, 0])
img[:, :, 1] = np.where(masks[:, :, idx], np.zeros((512, 512)),
img[:, :, 1])
img[:, :, 2] = np.where(masks[:, :, idx], np.zeros((512, 512)),
img[:, :, 2])
final_image = cv2.addWeighted(img, 1, final_img, 1, 0)
cv2.imshow('frame', cv2.resize(final_image, (1200, 800)))
except:
cv2.imshow('frame', cv2.resize(frame, (1200, 800)))
if cv2.waitKey(1) & 0xFF == ord('q'):
break
# After the loop release the cap object
def find_cars(img, scaler, model):
"""
This function searches for vehicles in an image with a pre-trained 'model'
and 'scaler'. It defines an image region of interest, search windows
at different scale and returns all search windows with a positive detection.
:param img: Image to search.
:param scaler: Per-column scaler fitted to feature vectors
:param model: Pre-trained linear SVC model.
:return: List of search windows with detected vehicles.
"""
# Set constants
colorspace = c.CSPACE # Color space
orient = c.ORIENT # Gradient orientation
pix_per_cell = c.PPC # Pixels per cell
cell_per_block = c.CPB # Cells per block
hog_channel = c.HCHNL # Hog channel
spatial_size = c.SPATIAL_SIZE
hist_bins = c.HIST_BINS
# Copy image
draw_img = np.copy(img)
search_img = convert_color(img, conv=c.CSPACE)
# Define scales for search windows
scales = [1.0, 1.5, 2.0, 2.5, 3.5]
detections = []
# Search image with search windows at different scale
for scale in scales:
# Get search windows for scale
window_list = slide_window(
img,
x_start_stop=[None, None],
y_start_stop=[c.YSTART, c.YSTART + (c.YSTOP - c.YSTART) * scale],
xy_window=(int(64 * scale), int(64 * scale)),
xy_overlap=(0.75, 0.75))
# Search windows for cars
for window in window_list:
# Extract window segment from image and convert color space
segment_tosearch = search_img[window[0][1]:window[1][1],
window[0][0]:window[1][0], :]
# Resize segement if scale > 1
if scale != 1:
imshape = segment_tosearch.shape
segment_tosearch = cv2.resize(
segment_tosearch,
(np.int(imshape[1] / scale), np.int(imshape[0] / scale)))
# Get color channels
ch1 = segment_tosearch[:, :, 0]
ch2 = segment_tosearch[:, :, 1]
ch3 = segment_tosearch[:, :, 2]
# Compute individual channel HOG features for the entire image
hog1 = get_hog_features(ch1,
orient,
pix_per_cell,
cell_per_block,
feature_vec=False).ravel()
hog2 = get_hog_features(ch2,
orient,
pix_per_cell,
cell_per_block,
feature_vec=False).ravel()
hog3 = get_hog_features(ch3,
orient,
pix_per_cell,
cell_per_block,
feature_vec=False).ravel()
# Stack HOG features for this patch
hog_features = np.hstack((hog1, hog2, hog3))
# Get color features
# spatial_features = get_color_bin_features(segment_tosearch)
# hist_features = get_color_hist_features(segment_tosearch)
# # Stack all features
# test_features = np.hstack((spatial_features, hist_features, hog_features)).reshape(1, -1)
test_features = hog_features.reshape(1, -1)
# Scale features and make a prediction
test_features = scaler.transform(test_features)
test_prediction = model.predict(test_features)
# If vehicle detected, add to detections
if test_prediction == 1:
detections.append(window)
return detections
def extract_features(imgs,
cspace=c.CSPACE,
spatial_size=(32, 32),
hist_bins=c.HIST_BINS,
orient=c.ORIENT,
pix_per_cell=c.PPC,
cell_per_block=c.CPB,
hog_channel='ALL',
spatial_feat=False,
hist_feat=False,
hog_feat=True):
# Create a list to append feature vectors to
features = []
# Iterate through the list of images
for file in imgs:
print(file)
file_features = []
# Read in each one by one
image = cv2.imread(file)
# image = mpimg.imread(file)
# image = image*255
# Convert color space of image
feature_image = convert_color(image, conv=cspace)
# Get color spacial features
if spatial_feat == True:
spatial_features = get_color_bin_features(feature_image,
size=spatial_size)
file_features.append(spatial_features)
# Get color histogram features
if hist_feat == True:
# Apply get_color_hist_features()
hist_features = get_color_hist_features(feature_image,
nbins=hist_bins)
file_features.append(hist_features)
# Get hog features for each channel
if hog_feat == True:
if hog_channel == 'ALL':
hog_features = []
for channel in range(feature_image.shape[2]):
hog_features.append(
get_hog_features(feature_image[:, :, channel],
orient,
pix_per_cell,
cell_per_block,
vis=False,
feature_vec=True))
hog_features = np.ravel(hog_features)
else:
hog_features = get_hog_features(feature_image[:, :,
hog_channel],
orient,
pix_per_cell,
cell_per_block,
vis=False,
feature_vec=True)
# Add hog features to image features
file_features.append(hog_features)
# Add image features to overall features list
features.append(np.concatenate(file_features))
return features
label_descriptions = json.load(f)
vid = cv2.VideoCapture(0)
vid.set(cv2.CAP_PROP_FRAME_WIDTH, 512)
vid.set(cv2.CAP_PROP_FRAME_HEIGHT, 512)
while (True):
ret, frame = vid.read()
try:
img, r = masker_np(frame)
masks, rois = convert_to_mask(img, r)
idx = list(r['class_ids']).index(2)
res = crop_by_id(img, masks, idx)
final_img = convert_color(img, res, [255, 105, 180])
img[:, :, 0] = np.where(masks[:, :, idx], np.zeros((512, 512)),
img[:, :, 0])
img[:, :, 1] = np.where(masks[:, :, idx], np.zeros((512, 512)),
img[:, :, 1])
img[:, :, 2] = np.where(masks[:, :, idx], np.zeros((512, 512)),
img[:, :, 2])
final_image = cv2.addWeighted(img, 1, final_img, 1, 0)
cv2.imshow('frame', final_image)
except:
cv2.imshow('frame', cv2.resize(frame, (512, 512)))
if cv2.waitKey(1) & 0xFF == ord('q'):
break
# After the loop release the cap object
def search_cars_with_option(self, img, scale, cells_per_step, ystart, ystop, conf_thresh):
"""
Detects car bboxes of image with given scale in region of img[ystart:ystop:,:,:]
:param img: input image
:param scale: window scale.
:param cells_per_step: cells per step.
:param ystart: y-range start.
:param ystop: y-range stop.
:param conf_thresh: classifier confidence threshold.
:return: list of (bbox, confidence)
"""
cvt_img = convert_color(img, self.P.color_space)
# Crop image on in y-region
ystart = 0 if ystart is None else ystart
ystop = img.shape[1] if ystop is None else ystop
cvt_img = cvt_img[ystart:ystop,:,:]
# Scale the image.
if scale != 1:
cvt_img = cv2.resize(cvt_img, (np.int(cvt_img.shape[1] / scale), np.int(cvt_img.shape[0] / scale)))
# Define blocks and steps as above
nxblocks = (cvt_img.shape[1] // self.P.pix_per_cell) - self.P.cell_per_block + 1
nyblocks = (cvt_img.shape[0] // self.P.pix_per_cell) - self.P.cell_per_block + 1
nblocks_per_window = (self.P.window_size[0] // self.P.pix_per_cell) - self.P.cell_per_block + 1
nxsteps = (nxblocks - nblocks_per_window) // cells_per_step + 1
nysteps = (nyblocks - nblocks_per_window) // cells_per_step + 1
# Compute individual channel HOG features for the entire image
hogs = []
for ch in range(cvt_img.shape[2]):
hogs.append(get_channel_hog_features(
img=cvt_img[:,:,ch], orient=self.P.orient,
pix_per_cell=self.P.pix_per_cell, cell_per_block=self.P.cell_per_block,
feature_vec=False, vis=False))
bbox_confs = []
for xb in range(nxsteps):
for yb in range(nysteps):
ypos = yb * cells_per_step
xpos = xb * cells_per_step
hog_features = []
for ch in range(cvt_img.shape[2]):
hog_features.append(
hogs[ch][ypos:ypos + nblocks_per_window, xpos:xpos + nblocks_per_window].ravel())
hog_features = np.hstack((hog_features[0], hog_features[1], hog_features[2]))
# Extract the image patch
xleft = xpos * self.P.pix_per_cell
ytop = ypos * self.P.pix_per_cell
subimg = cv2.resize(cvt_img[ytop:ytop + self.P.window_size[0],
xleft:xleft + self.P.window_size[0]],
self.P.window_size)
# Get spatial features
spatial_features = get_spatial_features(subimg, self.P.spatial_size)
# Get color features
color_features = get_color_features(subimg, size=self.P.window_size, nbins=self.P.color_nbins)
window_features = self.scaler.transform(np.hstack(
(spatial_features, color_features, hog_features)).reshape(1, -1))
if self.clf.predict(window_features) == 1:
xbox_left = np.int(xleft * scale)
ytop_draw = np.int(ytop * scale)
box_draw = np.int(self.P.window_size[0] * scale)
confidence = self.clf.decision_function(window_features)[0]
if confidence < conf_thresh:
# Only consider window with confidence score >= threshold.
continue
bbox = [(xbox_left, ytop_draw+ystart), (xbox_left+box_draw,ytop_draw+ystart+box_draw)]
bbox_conf = (bbox, confidence)
bbox_confs.append(bbox_conf)
return bbox_confs