Hey Product Hunt 👋

I built Visual Verifier around a simple problem I kept coming back to:

An image or video processing job can succeed while the produced media is still wrong.

An anonymizer can complete successfully but miss one frame. A plate blur can modify the video while failing to cover the plate that actually needed protection. A masking, redaction or watermarking pipeline can run without the intended transformation appearing everywhere it was required.

Most tools focus on transforming visual media. Visual Verifier is the independent test that runs afterward.

It checks processed images and videos frame by frame and can identify whether processing occurred, which frames were missed, whether reviewed regions were actually covered, how failures behave over time, and whether the result should pass or fail CI.

The bundled demo deliberately misses frames 4, 8 and 12:

pip install visual-verifier
visual-verifier demo

It runs locally - the media does not need to be uploaded.

Visual Verifier is also available as a reusable GitHub Marketplace Action, so it can become a quality gate after anonymization or other visual-processing pipelines.

I’d especially value feedback from people working with computer vision, privacy engineering, anonymization, video processing, QA or CI:

What visual transformation would you want Visual Verifier to verify next?