def main(im_path, desc_name):
print('[INFO] Preparing to extract features for images in \'' + im_path + '\'')
# track HOG feature vectors and corresponding images
features = {}
# image dimensions
width = 128
height = 64
# feature descriptor
print('[INFO] Using the ' + desc_name.upper() + ' feature descriptor')
if desc_name == 'hog':
descriptor = HOG()
# evaluate image files
print('[INFO] Processing images and computing features')
for filename in os.listdir(im_path):
if not filename.endswith('.jpg'):
continue
im = cv2.imread(im_path + filename, cv2.COLOR_BGR2GRAY)
# resize image
im = cv2.resize(im, (width,height))
# binarize using Otsu's method
im = cv2.threshold(im, 128, 255, cv2.THRESH_BINARY | cv2.THRESH_OTSU)[1]
im[im == 255] = 1
# thin using Zhang and Suen's method
#im = skeletonize(im)
#im = im.astype(np.uint8)
# compute features
v = descriptor.compute(im)
features[filename] = v
# save data
print('[INFO] Saving features and corresponding image name to \'features/' + desc_name + '_features.pickle\'')
with open('./features/' + desc_name + '_features.pickle', 'wb') as handle:
pickle.dump(features, handle)
# construct the argument parser and parse the command line arguments
ap = argparse.ArgumentParser()
ap.add_argument("-c",
"--conf",
required=True,
help="path to the configuration file")
args = vars(ap.parse_args())
# load the configuration file
conf = Conf(args["conf"])
# initialize the HOG descriptor along with the list of data and labels
hog = HOG(orientations=conf["orientations"],
pixelsPerCell=tuple(conf["pixels_per_cell"]),
cellsPerBlock=tuple(conf["cells_per_block"]),
normalize=conf["normalize"],
block_norm="L1")
data = []
labels = []
# grab the set of ground-truth images and select a percentage of them for training
trnPaths = list(paths.list_images(conf["image_dataset"]))
trnPaths = random.sample(trnPaths,
int(len(trnPaths) * conf["percent_gt_images"]))
print("[INFO] describing training ROIs...")
# setup the progress bar
widgets = [
"Extracting: ",
progressbar.Percentage(), " ",
help="path to the configuration file")
ap.add_argument("-i",
"--image",
required=True,
help="path to the image to be classified")
args = vars(ap.parse_args())
# load the configuration file
conf = Conf(args["conf"])
# load the classifier, then initialize the Histogram of Oriented Gradients descriptor
# and the object detector
model = pickle.loads(open(conf["classifier_path"], "rb").read())
hog = HOG(orientations=conf["orientations"],
pixelsPerCell=tuple(conf["pixels_per_cell"]),
cellsPerBlock=tuple(conf["cells_per_block"]),
normalize=conf["normalize"],
block_norm="L1")
od = ObjectDetector(model, hog)
# load the image and convert it to grayscale
image = cv2.imread(args["image"])
image = imutils.resize(image, width=min(260, image.shape[1]))
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
# detect objects in the image
(boxes, probs) = od.detect(gray,
conf["window_dim"],
winStep=conf["window_step"],
pyramidScale=conf["pyramid_scale"],
minProb=conf["min_probability"])
def HOG_extractor(image_path, features):
image = cv2.imread(image_path)
image = imutils.resize(image, width=1024)
hog = HOG(image, None)
tuple = [image_path, hog]
features.append(tuple)