wehub-resource-sync
146 lines
5.2 KiB
Python
146 lines
5.2 KiB
Python
import os
|
|
from textwrap import wrap
|
|
|
|
import cv2
|
|
import matplotlib.pyplot as plt
|
|
import numpy as np
|
|
|
|
import shap
|
|
|
|
|
|
def is_empty(path):
|
|
"""Function to check if folder at given path exists and is not empty.
|
|
|
|
Returns True if folder is empty or does not exist.
|
|
"""
|
|
empty = False
|
|
if os.path.exists(path) and not os.path.isfile(path):
|
|
# Checking if the directory is empty or not
|
|
if not os.listdir(path):
|
|
empty = True
|
|
print("'test_images' folder is empty. Please place images to be tested in this folder.")
|
|
else:
|
|
empty = True
|
|
print(
|
|
"There is no 'test_images' folder under current directory. Please create one and place images to be tested there."
|
|
)
|
|
return empty
|
|
|
|
|
|
def make_dir(path):
|
|
"""Function to create a new directory with given path or empty if it already exists."""
|
|
if not os.path.exists(path):
|
|
if not os.path.isfile(path):
|
|
# make directory if it does not exist
|
|
os.makedirs(path)
|
|
else:
|
|
print("Please give a valid folder path.")
|
|
else:
|
|
# Check if empty or not
|
|
if os.listdir(path):
|
|
# if exists, empty directory
|
|
for file in os.listdir(path):
|
|
os.remove(os.path.join(path, file))
|
|
|
|
|
|
def add_sample_images(path):
|
|
"""Function to add sample images from imagenet50 SHAP data in the given folder."""
|
|
X, _ = shap.datasets.imagenet50()
|
|
counter = 1
|
|
indexes_list = [25, 26, 30, 44]
|
|
for i, image in enumerate(X):
|
|
if i in indexes_list:
|
|
path_to_image = os.path.join(path, f"{counter}.jpg")
|
|
save_image(image, path_to_image)
|
|
counter += 1
|
|
|
|
|
|
def load_image(path_to_image):
|
|
"""Function to load image at given path and return numpy array of RGB float values."""
|
|
image = cv2.imread(path_to_image)
|
|
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
|
|
return np.array(image).astype("float")
|
|
|
|
|
|
def check_valid_image(path_to_image):
|
|
"""Function to check if a file has valid image extensions and return True if it does.
|
|
Note: Azure Cognitive Services only accepts below file formats.
|
|
"""
|
|
valid_extensions = (".png", ".jpg", ".jpeg", ".gif", ".bmp", ".jfif")
|
|
if path_to_image.endswith(valid_extensions):
|
|
return True
|
|
|
|
|
|
def save_image(array, path_to_image):
|
|
"""Function to save image(RGB values array) at given path (filename and location)."""
|
|
# saving array of RGB values as an image
|
|
image = np.array(array) / 255.0
|
|
plt.imsave(path_to_image, image)
|
|
|
|
|
|
def resize_image(path_to_image, reshaped_dir):
|
|
"""Function to resize given image retaining original aspect ratio and save in given directory 'reshaped_dir'.
|
|
Returns numpy array of resized image and path where resized file is saved.
|
|
|
|
Note
|
|
----
|
|
Azure COGS CV has size limit of < 4MB and min size of 50x50 for images.
|
|
Hence, large image files are being reshaped in code below to increase speed of SHAP explanations and run Azure COGS for image captions.
|
|
If image (pixel_size, pixel_size) is greater than 500 for either of the dimensions:
|
|
1 - image is resized to have max. 500 pixel size for the dimension > 500
|
|
2 - other dimension is resized retaining the original aspect ratio
|
|
|
|
"""
|
|
image = load_image(path_to_image)
|
|
|
|
# checking if either of (pixel_size, pixel_size) dimension is greater than 500.
|
|
reshaped_path = None
|
|
_, tail = os.path.split(path_to_image)
|
|
file_name = tail.split(".")[0]
|
|
max_pixels = 500
|
|
reshape = True
|
|
|
|
if image.shape[0] == image.shape[1] and image.shape[0] > 500:
|
|
new_dim = (max_pixels, max_pixels)
|
|
elif image.shape[0] > image.shape[1] and image.shape[0] > 500:
|
|
new_dim = (max_pixels, int(image.shape[1] * max_pixels / image.shape[0]))
|
|
elif image.shape[1] > image.shape[0] and image.shape[1] > 500:
|
|
new_dim = (int(image.shape[0] * max_pixels / image.shape[1]), max_pixels)
|
|
else:
|
|
reshape = False
|
|
|
|
# reshape image
|
|
if reshape:
|
|
# flipping axis for cv2 because cv2 uses width x height while numpy uses height x width
|
|
image = cv2.resize(image, dsize=(new_dim[1], new_dim[0]))
|
|
reshaped_path = os.path.join(reshaped_dir, file_name + ".png")
|
|
print("Reshaped image size:", image.shape)
|
|
save_image(image, reshaped_path)
|
|
image = np.array(image).astype("float")
|
|
|
|
return image, reshaped_path
|
|
|
|
|
|
def display_grid_plot(list_of_captions, list_of_images, max_columns=4, figsize=(20, 20)):
|
|
"""Function to display grid of images and their titles/captions."""
|
|
# load list of images
|
|
masked_images = []
|
|
for filename in list_of_images:
|
|
image = load_image(filename)
|
|
masked_images.append(image.astype(int))
|
|
|
|
# display grid plot with wrapping
|
|
fig = plt.figure(figsize=figsize)
|
|
column = 0
|
|
for i in range(len(masked_images)):
|
|
column += 1
|
|
# check for end of column and create a new figure
|
|
if column == max_columns + 1:
|
|
fig = plt.figure(figsize=figsize)
|
|
column = 1
|
|
fig.add_subplot(1, max_columns, column)
|
|
plt.imshow(masked_images[i])
|
|
plt.axis("off")
|
|
if len(list_of_captions) >= len(masked_images):
|
|
plt.title("\n".join(wrap(str(list_of_captions[i]), width=40)))
|