Home / Image redaction software

Image Redaction Software

The same detection that runs on video, applied to stills. Faces, plates, identity documents and on-screen text, across 187 image formats, thousands of files at a time.

BATCH — 2,418 OF 4,000 DONEDONEDONEDONEDONEDONEDONEDONEDONEQUEUEDQUEUEDQUEUEDQUEUEDQUEUEDQUEUED 187 image formats in, no conversion step

Why Teams Move

187

Image formats, out of 315 across all content types

Bulk

Hundreds of files per batch, queued and worked overnight

360°

Panoramas handled with projection-aware detection

What Makes Images Hard

It is never one image

It is four thousand, exported from a camera system nobody documented, and a per-file workflow is arithmetically impossible.

Quality and angle vary

Sideways, skewed, low light, photographed rather than scanned. The detector has to cope with what the archive actually holds.

One image discloses a lot

An ID card, driving licence or passport lying on a desk gives up every field in a single frame. Identity documents are a detected class in their own right for exactly that reason.

Mixed content

A scanned page that is really an image, carrying printed text, handwriting and a photograph together.

360 panoramas break flat tools

Treating an equirectangular image as a rectangle puts the mask in the wrong place. Detection converts between the projection and spherical coordinates.

Formats nobody planned for

The awkward ones come from legacy camera systems, scientific instruments and twenty-year-old archives — exactly the ones a tool supporting "JPG, PNG, TIFF, BMP and more" refuses.

Metadata survives redaction

Images carry GPS coordinates, device identifiers and author fields. Metadata is stripped or rewritten on export.

What You Can Redact

Faces and people

Persons, faces, heads and body parts

Licence plates

Detected and read as text

Vehicles

Cars, buses, trucks, bikes, boats, aircraft and trains

Weapons

Firearms and other weapons

Screens and devices

Monitors, laptops, phones and in-car terminals

Identity documents

ID cards, driving licences, passports and name tags

Text in the image

Recognised through OCR, including non-Latin scripts

Tattoos and street signs

Identifying marks and location clues

Image metadata

GPS coordinates, device identifiers, author fields

187 Image Formats

The platform carries 315 distinct extensions across five content types, and images are the largest group at 187.

That number does real work. The long tail is rendered through ImageMagick, so an obscure raster format previews in the browser without the software that created it.

DICOM medical imaging is handled separately, as imaging rather than as a generic image. It has its own page.

How It Works

01

Upload

Individually, as folders, or by API. No conversion step.

02

Detect

Automatically from the object classes you select, or draw regions directly.

03

Review

Correct, delete false positives, add what was missed.

04

Produce

Blur, pixelate or solid fill, with the original governed by a configured policy.

Where images share a layout — a recurring form, a standard ID scan — a template applies the same pattern across the batch.

Proof

Claims documents and checks

Text, tables and embedded images

Inspection photographs

Faces, plates, identity documents

Inspection video

Objects tracked across frames

Recorded calls

Spoken PII, one speaker at a time

An insurance claims operation runs redaction across all four formats as an automated stage in the claims-folder pipeline, rather than a task an adjuster performs. Detection rules are tuned to the data formats their own carriers send, so a recurring form or a standard ID scan is processed the same way every time.

Claims work is the clearest case for handling images alongside everything else. A document-only tool covers the statements and leaves the photographs, and the organisation ends up running two workflows and reconciling them by hand.

More case studies

Compared With Doing It by Hand

Photoshop or a viewer Redactor
One file at a time Hundreds per batch, queued
Draw every box yourself Detection proposes, you correct
A flat tool distorts 360 panoramas Projection-aware detection
Convert the awkward formats first 187 formats in, no conversion
Metadata left in the file Stripped on export

Where It Runs

Shared SaaS, dedicated cloud, your own private cloud, on-premises, hybrid, or fully air-gapped.

Deployment options

FAQ

Image Redaction Software questions, answered

Can it process thousands of images at once?

Yes. Bulk image redaction is queued work that runs unattended, and bulk is permissioned per format, so authorising an operator for images does not authorise them for video.

How many image formats are supported?

187, out of 315 extensions across all content types. The long tail renders through ImageMagick, so an obscure raster format from a legacy camera or instrument previews in the browser without the software that created it.

Does it handle 360-degree panoramas?

Yes. Detection converts between the equirectangular projection and spherical coordinates, so the mask lands where the object actually is. A tool treating a panorama as a flat rectangle puts the box in the wrong place.

Can it redact text inside an image?

Yes. Text in an image is recognised through OCR and redacted like any other text, including in non-Latin scripts.

What about identity documents?

ID cards, driving licences, passports and name tags are a detected class in their own right, because a document lying on a desk discloses every field on it at once.

Are the image files still identifying after redaction?

Embedded metadata — GPS coordinates, device identifiers, author fields — is stripped or rewritten on export alongside the content redaction.

Name Your Awkward Format

Tell us the extension and we will tell you whether it ingests, previews and redacts — or send us a batch and we will run it.