mchesterkadwell · April 16, 2017 21:16 · portableant · Apr 17, 2017
diff --git a/faceDetectionPython3-a.py b/faceDetectionPython3-a.py
 #!/usr/bin/env python

 from SPARQLWrapper import SPARQLWrapper, JSON
 import urllib.request, urllib.parse, urllib.error
 import os
 from PIL import Image
 import subprocess
 import cv2
 import argparse
 import time
 ### Add import math for ceiling function
 import math

 # Retrieve images from British Museum Research Space and perform montage and facial recognition
 # Daniel Pett 21/3/2017
 # British Museum content is under a CC-BY-SA-NC license
 ### Tested on Python 3.5.1 by Mary Chester-Kadwell

 __author__ = "Daniel Pett"
 __credits__ = ["Richard Wareham", "Ben O'Steen", "Matthew Vincent"]
 __license__ = 'MIT'
 __version__ = "1.0.1"
 __maintainer__ = "Daniel Pett"
 __email__ = "dpett@britishmuseum.org"


 def make_executable(path):
    """
    Make the file executable
    :param path:
    :return:
    """
    mode = os.stat(path).st_mode
    mode |= (mode & 0o444) >> 2  # copy R bits to X
    os.chmod(path, mode)


 # Function defined for resize and crop of an image
 def resize_and_crop(img_path, modified_path, size, crop_type='top'):
    """
    Resize and crop an image to fit the specified size.
    args:
        img_path: path for the image to resize.
        modified_path: path to save the modified image.
        size: (width, height) a tuple. Eg (300,300)
        crop_type: 3 options 'top', 'middle' or 'bottom'
    raises:
        Exception: if this script cannot open the file provided by img_path, then the image will not save
        ValueError: thrown if an invalid `crop_type` is provided as an argument
    """
    # Resizing is done in this order: if height is higher than width resize is vertical, default is horizontal resize
    img = Image.open(img_path)

    # Get current and desired ratio for the images

    ### Add conversion to int as Python 3 division operator '/' expects numbers
    ### Remove conversion to float as Python 3 division operator '/' is 'true' division and produces a float
    img_ratio = int(img.size[0]) / int(img.size[1])
    ratio = int(size[0]) / int(size[1])

    # As mentioned above, the image is scaled and cropped vertically or
    # horizontally depending on the ratio.
    ### Change division operator '/' to Python 3 floor division '//' throughout as Wand (ImageMagick) functions expect int not float

    try:
        if ratio > img_ratio:
            img = img.resize((size[0], size[0] * int(img.size[1]) // int(img.size[0])),
                             Image.ANTIALIAS)
            # Switch for position of crop
            if crop_type == 'top':
                box = (0, 0, img.size[0], size[1])
            elif crop_type == 'middle':
                box = (0, (img.size[1] - size[1]) // 2,
                       img.size[0], (img.size[1] + size[1]) // 2)
            elif crop_type == 'bottom':
                box = (0, img.size[1] - size[1], img.size[0], img.size[1])
            else:
                raise ValueError('Error detected: Option is not valid for crop type')
            img = img.crop(box)
        elif ratio < img_ratio:
            img = img.resize((size[1] * img.size[0] // img.size[1], size[1]),
                             Image.ANTIALIAS)
            # Switch for position of crop
            if crop_type == 'top':
                box = (0, 0, size[0], img.size[1])
            elif crop_type == 'middle':
                box = ((img.size[0] - size[0]) // 2, 0,
                       (img.size[0] + size[0]) // 2, img.size[1])
            elif crop_type == 'bottom':
                box = (img.size[0] - size[0], 0, img.size[0], img.size[1])
            else:
                raise ValueError('Error detected: Option is not valid for crop type')
            img = img.crop(box)
        else:
            img = img.resize((size[0], size[1]),
                             Image.ANTIALIAS)

    except Exception as ex:
        template = "An exception of type {0} occurred. Arguments:\n{1!r}"
        message = template.format(type(ex).__name__, ex.args)
        print(message)

    img.save(modified_path)


 def count_files(path, extension):
    """
    Count number of files of a specific extension
    :param path:
    :param extension:
    :return:
    """
    list_dir = []
    list_dir = os.listdir(path)
    count = 0
    for fn in list_dir:
        if fn.endswith(extension):
            # eg: '.jpg'
            count += 1
    return count

 ### Using the file containing the list of resized images appears to be unnecessary ?
 ### Refactored to just iterate through the images available in the directory
 ### This also has the benefit of solving another bug whereby the script looks for files.txt in the base directory
 ### even though it has actually been written into the 'resized' directory
 def create_montage():
    """
    Create the montage if at least one resized image exists with a try catch block
    :return:
    """

    if (count_files(paths[args.resized], ".jpg") > 0):
        print("Resized images exist")
        try:
            # This will produce multiple tiles for large results
            print("Now creating image montage of all retrieved (resized) images")
            subprocess.call("montage -border 0 -geometry 660x -tile 10x10 " +
                            args.resized + "/* " +
                            "/".join([args.montages, args.output]) + ".jpg",
                            shell=True)
            print("Now resizing image montage of all retrieved (resized) images")
            subprocess.call("convert " +
                            "/".join([args.montages, args.output]) + ".jpg -resize 750 " +
                            "/".join([args.montages, args.output]) + "_montage_750w.jpg",
                            shell=True)
            # This call makes a montage of the faces detected
            print("Now creating image montage of all faces detected in images")
            subprocess.call("montage -border 0 -geometry 660x -tile " + dims + " " +
                            args.faces + "/* " +
                            "/".join([args.montages, args.output]) + "Faces.jpg",
                            shell=True)
            print("Now resizing image montage of all faces detected in images")
            subprocess.call("convert " +
                            "/".join([args.montages, args.output]) + "Faces.jpg -resize 750 " +
                            "/".join([args.montages, args.output]) + "Faces_montage_750w.jpg",
                            shell=True)
        except:
            # The process failed
            raise ValueError("Montage generation failed")
    else:
        print("No resized images exist")


 if __name__ == "__main__":
    parser = argparse.ArgumentParser(
        description='A script for retrieving images from British Museum Research Space and '
                    'perform montage and facial recognition')

    parser.add_argument('-p', '--path', help='The path in which to run this script', required=True)
    # An example would be: --path '/Users/danielpett/githubProjects/scripts/'

    parser.add_argument('-d', '--download', help='The path in which to place downloaded original images', required=True)
    # An example would be: --download '/Users/danielpett/githubProjects/scripts/bmImages/'

    parser.add_argument('-r', '--resized', help='The path in which to place resized original images', required=True)
    # An example would be: --resized '/Users/danielpett/githubProjects/scripts/bmImagesResized/'

    parser.add_argument('-f', '--faces', help='The path in which to place images of detected faces', required=True)
    # An example would be: --faces '/Users/danielpett/githubProjects/scripts/facesDetected/'

    ### Add extra arg to specify folder for highlights otherwise it always goes into the base directory
    parser.add_argument('-i', '--highlights', help='The path in which to place original images with detected faces highlighted', required=True)
    # An example would be: --highlights '/Users/danielpett/githubProjects/scripts/bmImagesFaceHighlights/'

    parser.add_argument('-m', '--montages', help='The path in which to place montage images', required=True)
    # An example would be: --montages '/Users/danielpett/githubProjects/scripts/montages/'

    ### Add type=int as Wand (ImageMagick) functions must have type int
    parser.add_argument('-s', '--size', help='The resize dimensions', type=int, required=False, default=300)
    # An example would be: --resized '/Users/danielpett/githubProjects/scripts/bmimagesResized/'

    parser.add_argument('-o', '--output', help='The file name output for image magick', required=False,
                        default='britishMuseumImages')
    # An example would be 'britishMuseumPortraits'

    parser.add_argument('-t', '--template', help='The spaqrl query template to use', required=False, default='default')
    # An example would be 'default' as this is concatenated to default.txt

    parser.add_argument('-q', '--query', help='The spaqrl query string to use', required=False, default='bust')
    # An example would be 'bust'

    parser.add_argument('-e', '--endpoint', help='The spaqrl endpoint to use', required=False,
                        default='http://collection.britishmuseum.org/sparql')

    # Parse arguments
    args = parser.parse_args()

    # Set base path
    basePath = args.path

    # Define the base directories

    ### Add 'args.highlights' for highlights directory
    paths = {x: os.path.join(basePath, x) for x in
             [args.download, args.resized, args.montages, args.faces, 'opencv', args.highlights]}

    # Create them if they don't already exist
    for path in list(paths.values()):
        if not os.path.exists(path):
            os.makedirs(path)

    # Set up your sparql endpoint
    sparql = SPARQLWrapper(args.endpoint)

    # Read text file sparql query
    with open("sparql/" + args.template + ".txt", "r") as sparqlQuery:
        # Format the query string retrieved from the text file with simple replacement
        query = sparqlQuery.read().format(string=args.query)

    # Return the query for the user to see
    print("Your sparql query reads as: \n" + query)

    # Set your query
    sparql.setQuery(query)

    # Return the JSON triples
    sparql.setReturnFormat(JSON)
    results = sparql.query().convert()

    ### file.txt seems unnecessary ?
    ### so commented out
    # # Open the file for writing urls (this is for image magick)
    # listImages = open(os.path.join(paths[args.resized], "files.txt"), 'w')

    # Iterate over the results
    for result in results["results"]["bindings"]:
        image = result["image"]["value"]
        if os.path.isfile(os.path.join(paths[args.download], os.path.basename(image))):
            print("File already exists")
        else:
            path = os.path.join(paths[args.download], os.path.basename(image))
            urllib.request.urlretrieve(image, path)
            print("Image " + os.path.basename(image) + " downloaded")

    # ### Writing the resized image paths to file seems unnecessary ?
    # ### so commented out
    # for fn in os.listdir(paths[args.download]):
    #     if not fn.startswith('.'):
    #         listImages.write(os.path.join(paths[args.resized], os.path.basename(fn)) + "\n")
    #         print("Image path written to file")

    # Resize and crop the 'download' files into the 'resized' directory
    # Iterate through files and crop as required
    for fn in os.listdir(paths[args.download]):
        # Make sure file is not a hidden one etc
        if not fn.startswith('.') and os.path.isfile(os.path.join(paths[args.download], fn)):
            # Open the file checking if it is valid or not. It fails otherwise :-(
            try:
                if not os.path.exists(os.path.join(paths[args.resized], fn)):
                    resize_and_crop(os.path.join(paths[args.download], fn),
                                    os.path.join(paths[args.resized], fn),
                                    (args.size, args.size))
                    print(fn + " resized")
                else:
                    print("Resized file exists")
            except:
                pass

    # Amended to be relevant to the base path?
    cascPath = os.path.join(paths["opencv"], "haarcascade_frontalface_default.xml")
    # Check you have this file, if not get it
    if not os.path.isfile(cascPath):
        haar = "https://raw.githubusercontent.com/opencv/opencv/master/data/haarcascades/haarcascade_frontalface_default.xml"
        urllib.request.urlretrieve(haar, cascPath)

    # Create the haar cascade
    faceCascade = cv2.CascadeClassifier(cascPath)
    start = time.time()
    for fn in os.listdir(paths[args.download]):
        if not fn.startswith('.'):
            print("Detecting faces in " + os.path.join(paths[args.download], fn))
            image = cv2.imread(os.path.join(paths[args.download], fn))

            gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
            faces = faceCascade.detectMultiScale(
                gray,
                scaleFactor=1.1,
                minNeighbors=5,
                minSize=(150, 150),
                flags=cv2.CASCADE_SCALE_IMAGE
            )

            left = 10
            right = 10
            top = 10
            bottom = 10
            print("Found {0} faces within the image".format(len(faces)))

            if len(faces) > 0:
                for (x, y, w, h) in faces:
                    cv2.rectangle(image, (x, y), (x + w, y + h), (102, 0, 204), 2)
                    ### Add 'args.highlights' for highlights directory
                    rectName = os.path.join(paths[args.highlights], "high_{1}_{0}".format(str(fn), str(x)))
                    cv2.imwrite(rectName, image)
                    print("Written highlighted face " + rectName)

                    ### Image is acquired again from 'download' directory so that crop does not include rectangle
                    ### as drawn in the previous step
                    ### Side effect of this is to prevent crops happening off the edge of the image if there is more than
                    ### one face detected - previously this had been creating 'bad' empty files that needed to be
                    ### handled below by checking for filesize
                    ### Further side effect of this is previously unrealised detected faces have surfaced from the
                    ### demonstration query (1 of these is a new face, 1 is a duplicate face, 2 are not faces)
                    image = cv2.imread(os.path.join(paths[args.download], fn))
                    image = image[y - top:y + h + bottom, x - left:x + w + right]
                    filename = os.path.join(paths[args.faces], "cropped_{1}_{0}".format(str(fn), str(x)))
                    ### Move 'print("Written cropped face " + filename)' as the file is cropped close to the face below

                    if not os.path.exists(filename):
                        cv2.imwrite(filename, image)
                        ### Move line from above here as this is where the image is actually cropped close to the face
                        print("Written cropped face " + filename)
                        filesize = os.stat(filename).st_size
                        try:
                            if not filesize == 0:
                                resize_and_crop(filename, filename, (args.size, args.size), crop_type='middle')
                            else:
                                print(filename + " is likely to be broken.")
                                os.remove(filename)
                                print(filename + " has therefore been removed.")
                        except:
                            pass

    end = time.time()
    print("The time taken to process face detection was: " + "--- %s seconds ---" % (end - start))

    # ### Apparently duplicated and therefore unnecessary code from above ?
    # ### commented out
    # #Iterate through files and crop as required
    # for fn in os.listdir(paths[args.faces]):
    #     # Make sure file is not a hidden one etc
    #     if not fn.startswith('.') and os.path.isfile(os.path.join(paths[args.faces], fn)):
    #         # Open the file checking if it is valid or not. It fails otherwise :-(
    #         try:
    #             if not os.path.exists(os.path.join(paths[args.faces], fn)):
    #                 resize_and_crop(os.path.join(paths[args.faces], fn), os.path.join(paths[args.faces], fn), (args.size, args.size))
    #                 print(fn + " resized")
    #             else:
    #                 print("Resized file " + fn + " exists")
    #         except:
    #             pass

    a = count_files(paths[args.faces], ".jpg")
    print(str(a) + " faces were identified in total")
    ### Equation 'a/10' does not provide sufficient rows unless number of images is an exact multiple of 10
    ### Change equation to add math.ceil(), which returns smallest int >= input
    dims = "10x" + str(math.ceil(a / 10))
    print("The dimensions of the montage are " + dims)

    create_montage()
    print("Facial detection complete")
	#!/usr/bin/env python

	from SPARQLWrapper import SPARQLWrapper, JSON
	import urllib.request, urllib.parse, urllib.error
	import os
	from PIL import Image
	import subprocess
	import cv2
	import argparse
	import time
	### Add import math for ceiling function
	import math

	# Retrieve images from British Museum Research Space and perform montage and facial recognition
	# Daniel Pett 21/3/2017
	# British Museum content is under a CC-BY-SA-NC license
	### Tested on Python 3.5.1 by Mary Chester-Kadwell

	__author__ = "Daniel Pett"
	__credits__ = ["Richard Wareham", "Ben O'Steen", "Matthew Vincent"]
	__license__ = 'MIT'
	__version__ = "1.0.1"
	__maintainer__ = "Daniel Pett"
	__email__ = "dpett@britishmuseum.org"


	def make_executable(path):
	"""
	Make the file executable
	:param path:
	:return:
	"""
	mode = os.stat(path).st_mode
	mode \|= (mode & 0o444) >> 2 # copy R bits to X
	os.chmod(path, mode)


	# Function defined for resize and crop of an image
	def resize_and_crop(img_path, modified_path, size, crop_type='top'):
	"""
	Resize and crop an image to fit the specified size.
	args:
	img_path: path for the image to resize.
	modified_path: path to save the modified image.
	size: (width, height) a tuple. Eg (300,300)
	crop_type: 3 options 'top', 'middle' or 'bottom'
	raises:
	Exception: if this script cannot open the file provided by img_path, then the image will not save
	ValueError: thrown if an invalid `crop_type` is provided as an argument
	"""
	# Resizing is done in this order: if height is higher than width resize is vertical, default is horizontal resize
	img = Image.open(img_path)

	# Get current and desired ratio for the images

	### Add conversion to int as Python 3 division operator '/' expects numbers
	### Remove conversion to float as Python 3 division operator '/' is 'true' division and produces a float
	img_ratio = int(img.size[0]) / int(img.size[1])
	ratio = int(size[0]) / int(size[1])

	# As mentioned above, the image is scaled and cropped vertically or
	# horizontally depending on the ratio.
	### Change division operator '/' to Python 3 floor division '//' throughout as Wand (ImageMagick) functions expect int not float

	try:
	if ratio > img_ratio:
	img = img.resize((size[0], size[0] * int(img.size[1]) // int(img.size[0])),
	Image.ANTIALIAS)
	# Switch for position of crop
	if crop_type == 'top':
	box = (0, 0, img.size[0], size[1])
	elif crop_type == 'middle':
	box = (0, (img.size[1] - size[1]) // 2,
	img.size[0], (img.size[1] + size[1]) // 2)
	elif crop_type == 'bottom':
	box = (0, img.size[1] - size[1], img.size[0], img.size[1])
	else:
	raise ValueError('Error detected: Option is not valid for crop type')
	img = img.crop(box)
	elif ratio < img_ratio:
	img = img.resize((size[1] * img.size[0] // img.size[1], size[1]),
	Image.ANTIALIAS)
	# Switch for position of crop
	if crop_type == 'top':
	box = (0, 0, size[0], img.size[1])
	elif crop_type == 'middle':
	box = ((img.size[0] - size[0]) // 2, 0,
	(img.size[0] + size[0]) // 2, img.size[1])
	elif crop_type == 'bottom':
	box = (img.size[0] - size[0], 0, img.size[0], img.size[1])
	else:
	raise ValueError('Error detected: Option is not valid for crop type')
	img = img.crop(box)
	else:
	img = img.resize((size[0], size[1]),
	Image.ANTIALIAS)

	except Exception as ex:
	template = "An exception of type {0} occurred. Arguments:\n{1!r}"
	message = template.format(type(ex).__name__, ex.args)
	print(message)

	img.save(modified_path)


	def count_files(path, extension):
	"""
	Count number of files of a specific extension
	:param path:
	:param extension:
	:return:
	"""
	list_dir = []
	list_dir = os.listdir(path)
	count = 0
	for fn in list_dir:
	if fn.endswith(extension):
	# eg: '.jpg'
	count += 1
	return count

	### Using the file containing the list of resized images appears to be unnecessary ?
	### Refactored to just iterate through the images available in the directory
	### This also has the benefit of solving another bug whereby the script looks for files.txt in the base directory
	### even though it has actually been written into the 'resized' directory
	def create_montage():
	"""
	Create the montage if at least one resized image exists with a try catch block
	:return:
	"""

	if (count_files(paths[args.resized], ".jpg") > 0):
	print("Resized images exist")
	try:
	# This will produce multiple tiles for large results
	print("Now creating image montage of all retrieved (resized) images")
	subprocess.call("montage -border 0 -geometry 660x -tile 10x10 " +
	args.resized + "/* " +
	"/".join([args.montages, args.output]) + ".jpg",
	shell=True)
	print("Now resizing image montage of all retrieved (resized) images")
	subprocess.call("convert " +
	"/".join([args.montages, args.output]) + ".jpg -resize 750 " +
	"/".join([args.montages, args.output]) + "_montage_750w.jpg",
	shell=True)
	# This call makes a montage of the faces detected
	print("Now creating image montage of all faces detected in images")
	subprocess.call("montage -border 0 -geometry 660x -tile " + dims + " " +
	args.faces + "/* " +
	"/".join([args.montages, args.output]) + "Faces.jpg",
	shell=True)
	print("Now resizing image montage of all faces detected in images")
	subprocess.call("convert " +
	"/".join([args.montages, args.output]) + "Faces.jpg -resize 750 " +
	"/".join([args.montages, args.output]) + "Faces_montage_750w.jpg",
	shell=True)
	except:
	# The process failed
	raise ValueError("Montage generation failed")
	else:
	print("No resized images exist")


	if __name__ == "__main__":
	parser = argparse.ArgumentParser(
	description='A script for retrieving images from British Museum Research Space and '
	'perform montage and facial recognition')

	parser.add_argument('-p', '--path', help='The path in which to run this script', required=True)
	# An example would be: --path '/Users/danielpett/githubProjects/scripts/'

	parser.add_argument('-d', '--download', help='The path in which to place downloaded original images', required=True)
	# An example would be: --download '/Users/danielpett/githubProjects/scripts/bmImages/'

	parser.add_argument('-r', '--resized', help='The path in which to place resized original images', required=True)
	# An example would be: --resized '/Users/danielpett/githubProjects/scripts/bmImagesResized/'

	parser.add_argument('-f', '--faces', help='The path in which to place images of detected faces', required=True)
	# An example would be: --faces '/Users/danielpett/githubProjects/scripts/facesDetected/'

	### Add extra arg to specify folder for highlights otherwise it always goes into the base directory
	parser.add_argument('-i', '--highlights', help='The path in which to place original images with detected faces highlighted', required=True)
	# An example would be: --highlights '/Users/danielpett/githubProjects/scripts/bmImagesFaceHighlights/'

	parser.add_argument('-m', '--montages', help='The path in which to place montage images', required=True)
	# An example would be: --montages '/Users/danielpett/githubProjects/scripts/montages/'

	### Add type=int as Wand (ImageMagick) functions must have type int
	parser.add_argument('-s', '--size', help='The resize dimensions', type=int, required=False, default=300)
	# An example would be: --resized '/Users/danielpett/githubProjects/scripts/bmimagesResized/'

	parser.add_argument('-o', '--output', help='The file name output for image magick', required=False,
	default='britishMuseumImages')
	# An example would be 'britishMuseumPortraits'

	parser.add_argument('-t', '--template', help='The spaqrl query template to use', required=False, default='default')
	# An example would be 'default' as this is concatenated to default.txt

	parser.add_argument('-q', '--query', help='The spaqrl query string to use', required=False, default='bust')
	# An example would be 'bust'

	parser.add_argument('-e', '--endpoint', help='The spaqrl endpoint to use', required=False,
	default='http://collection.britishmuseum.org/sparql')

	# Parse arguments
	args = parser.parse_args()

	# Set base path
	basePath = args.path

	# Define the base directories

	### Add 'args.highlights' for highlights directory
	paths = {x: os.path.join(basePath, x) for x in
	[args.download, args.resized, args.montages, args.faces, 'opencv', args.highlights]}

	# Create them if they don't already exist
	for path in list(paths.values()):
	if not os.path.exists(path):
	os.makedirs(path)

	# Set up your sparql endpoint
	sparql = SPARQLWrapper(args.endpoint)

	# Read text file sparql query
	with open("sparql/" + args.template + ".txt", "r") as sparqlQuery:
	# Format the query string retrieved from the text file with simple replacement
	query = sparqlQuery.read().format(string=args.query)

	# Return the query for the user to see
	print("Your sparql query reads as: \n" + query)

	# Set your query
	sparql.setQuery(query)

	# Return the JSON triples
	sparql.setReturnFormat(JSON)
	results = sparql.query().convert()

	### file.txt seems unnecessary ?
	### so commented out
	# # Open the file for writing urls (this is for image magick)
	# listImages = open(os.path.join(paths[args.resized], "files.txt"), 'w')

	# Iterate over the results
	for result in results["results"]["bindings"]:
	image = result["image"]["value"]
	if os.path.isfile(os.path.join(paths[args.download], os.path.basename(image))):
	print("File already exists")
	else:
	path = os.path.join(paths[args.download], os.path.basename(image))
	urllib.request.urlretrieve(image, path)
	print("Image " + os.path.basename(image) + " downloaded")

	# ### Writing the resized image paths to file seems unnecessary ?
	# ### so commented out
	# for fn in os.listdir(paths[args.download]):
	# if not fn.startswith('.'):
	# listImages.write(os.path.join(paths[args.resized], os.path.basename(fn)) + "\n")
	# print("Image path written to file")

	# Resize and crop the 'download' files into the 'resized' directory
	# Iterate through files and crop as required
	for fn in os.listdir(paths[args.download]):
	# Make sure file is not a hidden one etc
	if not fn.startswith('.') and os.path.isfile(os.path.join(paths[args.download], fn)):
	# Open the file checking if it is valid or not. It fails otherwise :-(
	try:
	if not os.path.exists(os.path.join(paths[args.resized], fn)):
	resize_and_crop(os.path.join(paths[args.download], fn),
	os.path.join(paths[args.resized], fn),
	(args.size, args.size))
	print(fn + " resized")
	else:
	print("Resized file exists")
	except:
	pass

	# Amended to be relevant to the base path?
	cascPath = os.path.join(paths["opencv"], "haarcascade_frontalface_default.xml")
	# Check you have this file, if not get it
	if not os.path.isfile(cascPath):
	haar = "https://raw.githubusercontent.com/opencv/opencv/master/data/haarcascades/haarcascade_frontalface_default.xml"
	urllib.request.urlretrieve(haar, cascPath)

	# Create the haar cascade
	faceCascade = cv2.CascadeClassifier(cascPath)
	start = time.time()
	for fn in os.listdir(paths[args.download]):
	if not fn.startswith('.'):
	print("Detecting faces in " + os.path.join(paths[args.download], fn))
	image = cv2.imread(os.path.join(paths[args.download], fn))

	gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
	faces = faceCascade.detectMultiScale(
	gray,
	scaleFactor=1.1,
	minNeighbors=5,
	minSize=(150, 150),
	flags=cv2.CASCADE_SCALE_IMAGE
	)

	left = 10
	right = 10
	top = 10
	bottom = 10
	print("Found {0} faces within the image".format(len(faces)))

	if len(faces) > 0:
	for (x, y, w, h) in faces:
	cv2.rectangle(image, (x, y), (x + w, y + h), (102, 0, 204), 2)
	### Add 'args.highlights' for highlights directory
	rectName = os.path.join(paths[args.highlights], "high_{1}_{0}".format(str(fn), str(x)))
	cv2.imwrite(rectName, image)
	print("Written highlighted face " + rectName)

	### Image is acquired again from 'download' directory so that crop does not include rectangle
	### as drawn in the previous step
	### Side effect of this is to prevent crops happening off the edge of the image if there is more than
	### one face detected - previously this had been creating 'bad' empty files that needed to be
	### handled below by checking for filesize
	### Further side effect of this is previously unrealised detected faces have surfaced from the
	### demonstration query (1 of these is a new face, 1 is a duplicate face, 2 are not faces)
	image = cv2.imread(os.path.join(paths[args.download], fn))
	image = image[y - top:y + h + bottom, x - left:x + w + right]
	filename = os.path.join(paths[args.faces], "cropped_{1}_{0}".format(str(fn), str(x)))
	### Move 'print("Written cropped face " + filename)' as the file is cropped close to the face below

	if not os.path.exists(filename):
	cv2.imwrite(filename, image)
	### Move line from above here as this is where the image is actually cropped close to the face
	print("Written cropped face " + filename)
	filesize = os.stat(filename).st_size
	try:
	if not filesize == 0:
	resize_and_crop(filename, filename, (args.size, args.size), crop_type='middle')
	else:
	print(filename + " is likely to be broken.")
	os.remove(filename)
	print(filename + " has therefore been removed.")
	except:
	pass

	end = time.time()
	print("The time taken to process face detection was: " + "--- %s seconds ---" % (end - start))

	# ### Apparently duplicated and therefore unnecessary code from above ?
	# ### commented out
	# #Iterate through files and crop as required
	# for fn in os.listdir(paths[args.faces]):
	# # Make sure file is not a hidden one etc
	# if not fn.startswith('.') and os.path.isfile(os.path.join(paths[args.faces], fn)):
	# # Open the file checking if it is valid or not. It fails otherwise :-(
	# try:
	# if not os.path.exists(os.path.join(paths[args.faces], fn)):
	# resize_and_crop(os.path.join(paths[args.faces], fn), os.path.join(paths[args.faces], fn), (args.size, args.size))
	# print(fn + " resized")
	# else:
	# print("Resized file " + fn + " exists")
	# except:
	# pass

	a = count_files(paths[args.faces], ".jpg")
	print(str(a) + " faces were identified in total")
	### Equation 'a/10' does not provide sufficient rows unless number of images is an exact multiple of 10
	### Change equation to add math.ceil(), which returns smallest int >= input
	dims = "10x" + str(math.ceil(a / 10))
	print("The dimensions of the montage are " + dims)

	create_montage()
	print("Facial detection complete")
No results found