Videohash is a Python library for detecting near-duplicate videos (Perceptual Video Hashing). Any video input can be used to build a 64-bit equivalent hash value with this package. The hash-values for identical/near-duplicate videos are the same or similar, implying that if the video is enlarged (upscaled/downscaled), transcoded, slightly cropped, or black-bars added/removed, the hash-value should remain unchanged or not vary substantially.
- Every one second, a frame from the input video is extracted, the frames are shrunk to a 144x144 pixel square, a collage is constructed that contains all of the resized frames(square-shaped), the collage's wavelet hash value is computed, and the video hash value for the original input video is computed.
- Videohash cannot be used to verify whether one video is a part of another (video fingerprinting). If the video is reversed or rotated by a substantial angle (greater than 10 degrees), Videohash will not provide the same or similar hash result, but you can always reverse the video manually and generate the hash value for reversed video.
To use this software, you must have FFmpeg installed. Please read
how to install FFmpeg if you don't already know how.
- Using pip:
$ pip install videohash- Install directly from GitHub:
$ pip install git+https://github.com/akamhy/videohash.git- Generate videohash of a video directly from its URL or its path.
- Can be used to implement scalable Near Duplicate Video Retrieval.
- The end-user can see an image representation of the video.
- A videohash instance can be compared to a 64-bit stored hash, its hex representation, and other videohash instances.
- Faster than the old method of comparing each frame individually. The videohash package generates a single 64-bit video hash value, which saves a significant amount of database space. And the number of comparisons required drops dramatically.
>>> from videohash import VideoHash
>>> hash1 = VideoHash(url="https://www.youtube.com/watch?v=PapBjpzRhnA", download_worst=False) # video : Artemis I Hot Fire Test
>>> str(hash1) # str representation of VideoHash object (the output is video's videohash value)
'0b0011010000011111111011111111111110001111011110000000000000000000'
>>> hash1.hash # video hash value of the file, value is same as str(hash1)
'0b0011010000011111111011111111111110001111011110000000000000000000'
>>> hash1.hash_hex # hexadecimal representation of the videohash value
'0x341fefff8f780000'
>>> repr(hash1) # representation of VideoHash object
'VideoHash(hash=0b0011010000011111111011111111111110001111011110000000000000000000, hash_hex=0x341fefff8f780000, collage_path=/tmp/tmpvfr41629/temp_storage_dir/79c95zh4bq0s/collage/collage.jpg, bits_in_hash=64)'
>>> hash1.video_path # path of the downloaded video
'/tmp/tmpvfr41629/temp_storage_dir/79c95zh4bq0s/video/video.webm'
>>> hash1.storage_path # the storage directory
'/tmp/tmpvfr41629/temp_storage_dir/79c95zh4bq0s/'
>>> hash1.collage_path # path of the generated collage, the wavelet hash of this special collage is videohash value of the input video
'/tmp/tmpvfr41629/temp_storage_dir/79c95zh4bq0s/collage/collage.jpg'
>>> hash1.delete_storage_path() # To delete the storage path, deleting it will also delete the collage, extracted frames, and the downloaded video.
>>> hash1.bits_in_hash # how many bits in the hash, always 64 (a constant)
64
>>> len(hash1) # length of the hash value string, 64(no of bits in hash) + 2(prefix '0b')
66
>>> hash2 = VideoHash(url="https://raw.githubusercontent.com/akamhy/videohash/main/assets/rocket.mkv") # video : Artemis I Hot Fire Test, yes same as hash1(downscaled)
>>> hash2.hash
'0b0011010000011111111011111111111110001111011110000000000000000000'
>>> hash2.hash_hex
'0x341fefff8f780000'
>>> hash1.hash_hex
'0x341fefff8f780000'
>>> hash1 - hash2 # videohash objects support application of '-' operator on them. The other value must be a string (prefixed with '0x' or '0b') or another VideoHash object
0
>>> hash2 - "0x341fefff8f780000"
0
>>> hash1 - "0b0011010000011111111011111111111110001111011110000000000000000000"
0
>>> hash1 - "0b1111111111111111111111111111111111111111111111111111111111111111"
32
>>> hash1 == hash2 # videohash objects support application of '==' operator on them. The other value must be a string (prefixed with '0x' or '0b') or another VideoHash object.
True
>>> hash1 == "0b0011010000011111111011111111111110001111011110000000000000000000"
True
>>> hash1 != hash2 # videohash objects support application of '!=' operator on them. The other value must be a string (prefixed with '0x' or '0b') or another VideoHash object.
False
>>> hash3 = VideoHash(path="/home/akamhy/Downloads/rocket.mkv") # video : Artemis I Hot Fire Test, yes same as hash2 (downloaded locally)
>>> hash3.hash_hex
'0x341fefff8f780000'
>>> hash3.hash
'0b0011010000011111111011111111111110001111011110000000000000000000'
>>> hash3 - hash2
0
>>> hash3 == hash1
False
>>> hash3 == hash2
True
>>> hash4 = VideoHash(url="https://www.youtube.com/watch?v=_T8cn2J13-4") # video : How We Are Going to the Moon - 4K, a completely different video from the first 3 videos
>>> hash4.hash_hex
'0x7cffff000000eff0'
>>> hash4 - "0x7cffff000000eff0"
0
>>> hash4.hash
'0b0111110011111111111111110000000000000000000000001110111111110000'
>>> hash4 - "0b0111110011111111111111110000000000000000000000001110111111110000"
0
>>> hash4 == hash3
False
>>> hash4 - hash2
34
>>> hash4 != hash2
True
>>> hash4 - "0b0011010000011111111011111111111110001111011110000000000000000000"
34
>>>Run the above code @ https://replit.com/@akamhy/videohash-usage-2xx-example-code-for-video-hashing#main.py
Wiki/Extended Usage/Docs : https://github.com/akamhy/videohash/wiki
Released under the MIT License. See license for details.
Videos are from NASA and are in the public domain.
NASA videos are in the public domain. NASA copyright policy states that "NASA material is not protected by copyright unless noted".