Anonymize without losing the record
A face you don’t need to publish is a face you shouldn’t ship. Blurring and controlled redaction let you move the analysis forward while protecting the people in it.
Protecting subjects is part of the review
Case media often contains faces of people with no stake in the outcome: bystanders, neighbors, minors who simply happened to be there. Publishing or sharing a raw export is an implicit decision to include them — and most workflows never make that decision deliberately. Privacy is not a footnote you add when someone complains; it is a property of the artifact you hand over.
DawaImg is free, runs entirely offline, and ships with no cloud, no telemetry, and no accounts. That means the people in your media stay on your machine by default — but the moment you export, you decide what else leaves. That decision, made explicitly, is where a defensible privacy workflow begins.
Blurring, redaction, and presets
DawaImg builds anonymization into the same review surface as detection, not into a separate graphics step. Detected faces can be blurred in place so the context — the scene, the frame, the sequence — stays intact while the identity does not. For situations where even a blurred face is too much, a redaction export option covers the harder cases, including a minor-redaction option for exports where a minor’s face must not ship in any form.
Working with presets keeps the output consistent across a project. Configure once what level of anonymization a given artifact needs — blurred for internal review, redacted for external sharing — and apply that same rule to every face it detects. Consistency is what makes the process auditable later, rather than a judgment call that varies from frame to frame.
Build it on your machine
Free, offline, with built-in anonymization. Windows & macOS.
Controlled exports for a narrow audience
The safest output is the smallest one that still supports the claim you are making. A redacted export for counsel or an editor contains the sequence relationships and timestamps that matter, without carrying unneeded biometric detail. Because DawaImg runs offline (as of v39.0), producing these variants never requires uploading the media anywhere — the source stays on disk, and only the specific artifact you choose to share actually leaves.
Practically, that means a tiered export habit: full-res originals stay private, a blurred variant supports internal discussion, and a redacted variant is the outermost boundary. Each tier is narrower than the last, and each one is produced from the same local review, so nothing is lost analytically along the way.
A defensible privacy workflow
Put together, the workflow is unglamorous and repeatable: review the media locally, document which subjects are in scope, pick an anonymization level per audience, and export only what that audience needs. Every step runs on your machine, and none of them require a network, an account, or a third party seeing the raw data. When someone later asks what you shared with whom — and what you deliberately did not — the answer is the export log, not a memory.
That is what privacy-first review is: not hiding the analysis, but deciding, artifact by artifact, exactly how much of the image it is reasonable for the world to see.