Private fingerprint protection

Protect your videos from unauthorized upload.

Create a TMK+PDQF perceptual fingerprint from an original video, or upload a .tmk fingerprint you generated yourself. Protected fingerprints are checked against future video submissions before normal processing begins.

Analyze an original video

Analyse an original video and our tool will generate its TMK+PDQF fingerprint.

Drop your videos here

or select them from your device
Multiple videos can be submitted at once

Choose videos

Your submitted videos remain private.

Videos are analyzed locally and securely within your browser. The original video file is never uploaded, transferred, stored, or sent to DP-Vids servers. During the analysis, a privacy-preserving TMK fingerprint is generated from the video. Only this fingerprint is submitted to our private content-protection system and retained as reference data. The TMK fingerprint allows our system to identify matching or substantially similar copies, including videos that may have been re-encoded, shortened, cropped, or otherwise modified, without requiring us to store the original protected video.

Upload an existing .tmk fingerprint

If you generated the fingerprint yourself, send only the .tmk file. The original video does not need to be uploaded here.

Drop your .tmk files here

or choose one or more TMK fingerprints from your device

Choose TMK files

Only submit fingerprints for content you are authorized to protect.

The imported TMK file is used as a private matching reference for future submissions.

What is TMK+PDQF?

TMK+PDQF is Meta's open-source perceptual video fingerprinting technology. Instead of relying on a normal file checksum, it analyzes visual characteristics across the video and creates a compact .tmk fingerprint. This makes it possible to recognize videos that are visually very similar even when the file itself has been re-encoded or has received minor visual changes.

A TMK fingerprint is not the video itself and cannot be played back as a video. When a new video is submitted, our protection system generates a fingerprint for that submission and compares similarity against protected TMK references. A sufficiently strong match can be rejected before normal video processing begins.

1. Original videoThe video is visually analyzed frame by frame.

2. TMK+PDQFA compact perceptual fingerprint is created.

3. Future uploadNew submissions receive their own fingerprint.

4. Similarity checkStrong matches can be automatically rejected.

Perceptual, not a file checksum TMK is designed to compare visual similarity, so two files do not have to be byte-for-byte identical to be related.

Compact private reference The resulting .tmk is a fixed-size binary signature used for comparison rather than a viewable copy of the video.

Best for full-video similarity TMK was primarily designed for complete or nearly complete videos and can tolerate some minor edits such as watermarking.

Not perfect for every edit Very short clips or heavily modified versions may be harder to identify, so fingerprint matching is an additional protection layer rather than an absolute guarantee.

Prefer to create the fingerprint yourself?

You can generate TMK fingerprints locally on your own computer using Meta's official open-source ThreatExchange TMK tools. This allows you to keep the original video on your own system and send us only the resulting .tmk file.

# After installing FFmpeg and downloading Meta ThreatExchange: cd ThreatExchange/tmk/cpp make # Generate a TMK fingerprint: ./tmk-hash-video -f /usr/bin/ffmpeg -i /path/to/video.mp4 -d .

View official TMK+PDQF documentation on GitHub ↗

Select your originals Choose one or more original video files that you want analyzed for the protection library.

Video analysis Our private system analyzes the video and generates a TMK+PDQF perceptual fingerprint.

Protected reference The TMK fingerprint can be compared directly with future submissions to identify protected material.