print("[INFO] computing hashes for needles...")
needle_paths = list(paths.list_images(args["needles"]))
# Dictionary containing needle images and their correspondent haystack image path
results = {}
# Loop over the needle paths
for p in needle_paths:
# Load the image from disk
image = cv2.imread(p)
# If the image is None then we could not load it from disk (so skip it)
if image is None:
continue
# Convert the image to grayscale and compute the hash
image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
imageHash = dhash(image)
# Grab all image paths that match the hash
matched_paths = hash_dict.get(imageHash, [])
# Save the results
results[p] = matched_paths
# Display results found
print("[INFO] input images and their correspondents:")
# Loop over each subdirectory and display it
for key in results:
print('Image: {}'.format(key))
if len(results[key]) > 0:
for img in results[key]:
print('-> {}'.format(img))
else:
print('-> No similar images found.')
args = vars(ap.parse_args())
# Grab the paths to images
print("[INFO] Computing hashes for images...")
images_paths1 = list(paths.list_images(args["compare1"]))
images_paths2 = list(paths.list_images(args["compare2"]))
# Remove the `\` character from any filenames containing a space
# (assuming you're executing the code on a Unix machine)
if sys.platform != "win32":
images_paths1 = [p.replace("\\", "") for p in images_paths1]
images_paths2 = [p.replace("\\", "") for p in images_paths2]
# Load the input query image and compute hash
query_image = cv2.imread(args["query"])
query_hash = convert_hash(dhash(query_image))
query_image_resized = cv2.resize(query_image, (160, 160))
# Results
results1 = {'hashing': [], 'resnet': []}
results2 = {'hashing': [], 'resnet': []}
# Loop over the images paths (first folder)
for (i, p) in enumerate(images_paths1):
# Load the image from disk
print("[INFO] Folder #1: Processing image {}/{}".format(
i + 1, len(images_paths1)))
image = cv2.imread(p)
# If the image is None then we could not load it from disk (so skip it)
if image is None:
continue
images_paths = [p.replace("\\", "") for p in images_paths]
# Initialize the dictionary that will map the image hash to
# corresponding image
hash_dict = {}
# Loop over the images paths
for (i, p) in enumerate(images_paths):
# Load the image from disk
print("[INFO] processing image {}/{}".format(i + 1, len(images_paths)))
image = cv2.imread(p)
# If the image is None then we could not load it from disk (so skip it)
if image is None:
continue
# Compute the hash
h = convert_hash(dhash(image))
# Update the dictionary
list_images = hash_dict.get(h, [])
list_images.append(p)
hash_dict[h] = list_images
# Save dictionary to database
hash_filename = database_path + ".dict.pickle"
with open(hash_filename, "wb") as f:
pickle.dump(hash_dict, f)
print("[INFO] Hash data saved: " + hash_filename)
print("[INFO] Generating VPTree...")
tree = vptree.VPTree(list(hash_dict.keys()), hamming)
tree_filename = database_path + ".tree.pickle"
default=10,
help="Maximum hamming distance")
args = vars(ap.parse_args())
# Load the VP-Tree and hashes dictionary
print("[INFO] Loading VP-Tree and hashes...")
tree = pickle.loads(open(args["data"] + ".tree.pickle", "rb").read())
hashes = pickle.loads(open(args["data"] + ".dict.pickle", "rb").read())
# Load the input query image
image = cv2.imread(args["query"])
cv2.imshow("Query image", image)
# Compute the hash for the query image, then convert it
queryHash = convert_hash(dhash(image))
# Perform the search
print("[INFO] Performing search...")
start = time.time()
results = tree.get_all_in_range(queryHash, args["distance"])
results = sorted(results)
end = time.time()
print("[INFO] Search took {} seconds".format(end - start))
# Loop over the results
for (d, h) in results:
# Grab all image paths in our dataset with the same hash
result_paths = hashes.get(h, [])
print("[INFO] {} total image(s) with d: {}, h: {}".format(
len(result_paths), d, h))
'hashing': [],
'resnet': []
}
}, {
'filename': args["query22"],
'input': cv2.imread(args["query22"]),
'results': {
'hashing': [],
'resnet': []
}
}]]
# Loop over queries
for (i, p) in enumerate(queries):
# Loop over the images paths (first folder)
query_hash1 = convert_hash(dhash(p[0]['input']))
query_hash2 = convert_hash(dhash(p[1]['input']))
query_image1_resized = cv2.resize(p[0]['input'], (160, 160))
query_image2_resized = cv2.resize(p[1]['input'], (160, 160))
results1 = p[0]['results']
results2 = p[1]['results']
for (j, q) in enumerate(images_paths):
# Load the image from disk
print("[INFO] Query #{}: Processing image {}/{}".format(
i + 1, j + 1, len(images_paths)))
image = cv2.imread(q)
# If the image is None then we could not load it from disk (so skip it)
if image is None:
continue
# Compute the hash and difference hash
h = convert_hash(dhash(image))