Watermark detecting

How detection works

Say a copy of a video has leaked. Detection works on that recording alone: no reference copy of the original is needed (blind detection).

1. DoveRunner analyzes the frames of the leaked video.
2. It extracts the hidden watermark pattern.
3. It decrypts the pattern with the key used at insertion.
4. It matches the result against your session records, so you can see which session the leak came from.

Before you start

The video must meet these requirements, or detection fails:

• Required: one continuous recording of at least 5 minutes
• Required: at least 480p at 1 Mbps (720p or higher is recommended)
• Recommended: a steady, fixed recording with no shake
• Recommended: no buffering or freezing in the recording

Preparing the input video

Submit the recording as a file upload or as a link. A link can be:

• A direct-download URL from cloud storage, or a large-file transfer service link
• For live events, a non-DRM HLS stream

The link must start downloading with no login, manual confirmation, or interactive redirect. For Google Drive, set the file to open without signing in; a regular share link won't work.

1. Open Detection Service

In DoveRunner Console, go to Forensic Watermarking > Detection Service. The list shows every request you've made. Select + Request to create a new one.

You can also request detection with the Detection API. This lesson uses the Console.

2. Request detection

1. Choose the request type, File or URL.
2. Enter a Content Title: any name that helps you recognize the request.
3. Upload the file or paste the direct-download link.
4. If the video was recorded from the DoveRunner DEMO page instead of your own content, check Video captured from DoveRunner DEMO Page.

Advanced options

These are optional details about where the leak was found:

• Piracy URL: required if you add advanced options
• IP address and location: optional
• VPN Detected: switch on if the source used a VPN

Select Register to submit the request.

3. Follow the status

The request appears in the list and moves through these statuses:

Submitted > Downloading > Detecting > Completed

• Canceled: you stopped the request
• Error: the video didn't meet the requirements
• Failed: detection didn't succeed even though the video met them

4. Read the result

Select the content title to open the detail page.

• Detection File Information: detection ID, file name, video link, and SHA-256 hash
• Detection Result: status, the extracted watermark token, and the decoded payload data, which is the session information you embedded. If something went wrong, the error code and message appear here too.
• Detection Location and Source Information: the details you entered when you registered
• Detection Log: each processing step with a timestamp, from FD001 to FD004. You can download it.

Select Export Report to keep a record.

If detection can't extract the watermark

Trial accounts can make up to 2 detection requests. If the video is too distorted for automatic detection, send the original video or snapshots to the support team. They can run a manual extraction instead.

Course recap

1. Preprocessing: prepared your content as two variants (A/B)
2. Embedding: composed a unique stream for every viewer at playback time
3. Detecting: traces a leaked copy back to its session

See the Watermark Detecting guide for details.