def clusters(args):
db = decode(args.db)
imgs = list(db.images)
batches = imgs if args.num_threads == 1 else _split(imgs, args.num_threads)
pool = ThreadPool(args.num_threads)
batch_clusters = pool.map(
functools.partial(_get_neighbours, db.tree, args.num_neighbours,
args.min_distance), batches)
pool.close()
pool.join()
clusters_list = [item for sublist in batch_clusters for item in sublist]
clusters = {}
for i in range(0, len(imgs)):
if len(clusters_list[i]) != 0:
clusters[imgs[i].path] = clusters_list[i]
for img_path, nearest in clusters.items():
print('image: {}'.format(img_path))
print('\tpath\tdistance')
for n in nearest:
print('\t{}\t{}'.format(n[0], n[1]))
print()
def search_by_distance(args):
db = decode(args.db)
query = image.Image(args.query, db.hash_type, db.hash_size)
out = db.tree.get_within_distance(query, args.max_distance)
if len(out) == 0:
return
for img in out:
if os.path.abspath(args.query) == img.path:
continue
print('path {} distance: {}'.format(img.path, query.distance(img)))
def remove(args):
db = decode(args.db)
path = os.path.abspath(args.image)
img = image.Image(path, db.hash_type, db.hash_size)
if path in db.image_hashes:
if db.tree.remove(img):
del db.image_hashes[path]
db.encode(args.db)
else:
raise ValueError('path is in dict but not in tree')
def search_nearest(args):
db = decode(args.db)
query = image.Image(args.query, db.hash_type, db.hash_size)
out = db.tree.get_nearest_neighbours(query, args.num_neighbours,
args.max_results)
if len(out) == 0:
return
for img in out:
if os.path.abspath(args.query) == img.path:
continue
print('path {} distance: {}'.format(img.path, query.distance(img)))
def rebuild(args):
db = decode(args.db)
hash_size = args.hash_size if args.hash_size != 0 else db.hash_size
hash_type = args.hash_type if len(args.hash_type) != 0 else db.hash_type
imgs = []
files = []
for path in db.image_hashes:
img = image.Image(path, hash_type, hash_size)
imgs.append(img)
files.append(img)
tree = VPTree(distance_fn=image.distance_fn)
tree.add_list(imgs)
db = DB(tree, files, hash_type, hash_size)
db.encode(args.db)
def update(args):
db = decode(args.db)
walkdir(
args.dir, args.recursive, lambda path: update_func(
path, db.tree, db.image_hashes, db.hash_type, db.hash_size))
db.encode(args.db)
def add(args):
db = decode(args.db)
img = image.Image(args.image, db.hash_type, db.hash_size)
db.tree.add(img)
db.image_hashes.add(args.image)
db.encode(args.db)