Perception 0.9.0

Perception provides flexible, well-documented, and comprehensively tested tooling for perceptual hashing research, development, and production use.

perception

perception provides flexible, well-documented, and comprehensively tested tooling for perceptual hashing research, development, and production use. See the documentation for details.

Background

perception was initially developed at Thorn as part of our work to eliminate child sexual abuse material from the internet. For more information on the issue, check out our CEO's TED talk.

Getting Started

Installation

pip install perception

Optional extras

perception provides optional extras for additional functionality:

• approximate-deduplication – FAISS-based approximate-nearest-neighbor deduplication and graph community/clique detection (used by perception.approximate_deduplication and perception.local_descriptor_deduplication)
• benchmarking – tools for benchmarking perceptual hashes
• matching – async matching utilities
• pdq – Facebook's PDQ hash support

Note for benchmarking extra users: The benchmarking extra depends on albumentations, which in turn requires opencv-python-headless. However, perception already depends on opencv-contrib-python-headless (needed for contrib modules such as cv2.img_hash and cv2.SIFT_create). Installing both OpenCV distributions simultaneously causes file-level conflicts.

If you are using uv, this is handled automatically:

uv pip install "perception[benchmarking]"

If you are using plain pip, install the extra and then force-reinstall the contrib variant to remove the conflicting headless package:

pip install "perception[benchmarking]"
pip install --force-reinstall --no-deps opencv-contrib-python-headless

Hashing

Hashing with different functions is simple with perception.

from perception import hashers

file1, file2 = 'test1.jpg', 'test2.jpg'
hasher = hashers.PHash()
hash1, hash2 = hasher.compute(file1), hasher.compute(file2)
distance = hasher.compute_distance(hash1, hash2)

Examples

See below for end-to-end examples for common use cases for perceptual hashes.

• Detecting child sexual abuse material
• Deduplicating media
• Benchmarking perceptual hashes

Supported Hashing Algorithms

perception currently ships with:

• pHash (DCT hash) (perception.hashers.PHash)
• Facebook's PDQ Hash (perception.hashers.PDQ)
• dHash (difference hash) (perception.hashers.DHash)
• aHash (average hash) (perception.hashers.AverageHash)
• Marr-Hildreth (perception.hashers.MarrHildreth)
• Color Moment (perception.hashers.ColorMoment)
• Block Mean (perception.hashers.BlockMean)
• wHash (wavelet hash) (perception.hashers.WaveletHash)

Contributing

To work on the project, start by doing the following.

# Install local dependencies for code completion,
# testing, and linting.
make init

To do a (close to) comprehensive check before committing code, use make precommit.

To implement new features, please first file an issue proposing your change for discussion.

To report problems, please file an issue with sample code, expected results, actual results, and a complete traceback.

Alternatives

There are other packages worth checking out to see if they meet your needs for perceptual hashing. Here are some examples.

• dedupe
• imagededup
• ImageHash
• PhotoHash