def test_scene_detection_batches():
hasher = hashers.SimpleSceneDetection(base_hasher=hashers.TMKL1(
frames_per_second=30,
frame_hasher=hashers.PHashU8(),
norm=None,
distance_metric='euclidean'),
max_scene_length=10)
hashes_v2s = hasher.compute('perception/testing/videos/v2s.mov',
errors='raise')
hashes_batches = []
frame_count = 0
for batch in hasher.compute_batches('perception/testing/videos/v2s.mov',
batch_size=1):
for hash_string, frames in batch:
hashes_batches.append(hash_string)
frame_count += len(frames)
# Ensure we get the same hashes whether using compute or compute_batches
assert len(hashes_batches) == len(hashes_v2s)
assert all(h1 == h2 for h1, h2 in zip(hashes_batches, hashes_v2s))
expected_frame_count = 0
for _, _, _ in hashers.tools.read_video(
'perception/testing/videos/v2s.mov', frames_per_second=30):
expected_frame_count += 1
# Ensure all frames were accounted for in scene detection
assert expected_frame_count == frame_count
def test_scene_detection():
hasher = hashers.SimpleSceneDetection(base_hasher=hashers.TMKL1(
frames_per_second=30,
frame_hasher=hashers.PHashU8(),
norm=None,
distance_metric='euclidean'),
max_scene_length=10)
assert len(
hasher.compute('perception/testing/videos/v1.m4v',
errors='raise')) == 1
assert len(
hasher.compute('perception/testing/videos/v2.m4v',
errors='raise')) == 1
hashes_v2s = hasher.compute('perception/testing/videos/v2s.mov',
errors='raise')
assert len(hashes_v2s) == 2
def test_deduplicate_u8():
# This test verifies that extensions.compute_euclidean_pairwise_duplicates
# works properly.
directory = tempfile.TemporaryDirectory()
original = testing.DEFAULT_TEST_IMAGES[0]
duplicate = os.path.join(directory.name, 'image1.jpg')
shutil.copy(original, duplicate)
pairs = tools.deduplicate(files=[
testing.DEFAULT_TEST_IMAGES[0], testing.DEFAULT_TEST_IMAGES[1],
duplicate
],
hashers=[(hashers.PHashU8(hash_size=16), 10)])
assert len(pairs) == 1
file1, file2 = pairs[0]
assert ((file1 == duplicate) and
(file2 == original)) or ((file1 == original) and
(file2 == duplicate))
def test_deduplicate_hashes_multiple():
# This test verifies that deduplicate_hashes functions properly
# when there is more than one hash for a file.
directory = tempfile.TemporaryDirectory()
original = testing.DEFAULT_TEST_IMAGES[0]
duplicate = os.path.join(directory.name, 'image1.jpg')
hasher = hashers.PHashU8(hash_size=16)
shutil.copy(original, duplicate)
hashes = [
(0, hasher.compute(original)),
(1, hasher.compute(duplicate)),
(1, hasher.compute(duplicate)),
(1, hasher.compute(duplicate)),
(2, hasher.compute(testing.DEFAULT_TEST_IMAGES[1])),
]
pairs = tools.deduplicate_hashes(hashes=hashes,
threshold=10,
hash_format='base64',
hash_length=hasher.hash_length,
distance_metric='euclidean',
hash_dtype='uint8')
assert len(pairs) == 1
file1, file2 = pairs[0]
assert ((file1 == 0) and (file2 == 1)) or ((file1 == 1) and (file2 == 0))