# Blur People and Objects in a Photo Automatically

> Detect people, faces, plates and screens with an object-detection model in your browser, then blur or black them out.

This tool runs an object-detection model in your browser to find people, vehicles and screens in a photograph, then blurs, pixelates or blacks out each detected region. The detection and the editing both happen on your device, which matters for redaction specifically — a tool that uploads the image you are trying to anonymise has already defeated the purpose.

**URL:** https://convertto.tech/t/ai-photo-redactor
**Category:** On-Device AI (https://convertto.tech/c/local-ai-tools)
**Privacy:** Runs entirely in the browser; no upload
**Cost:** Free, no sign-up
**Last updated:** 2026-08-01

## Key facts

- **Runs on:** WebGPU where available, WebAssembly otherwise
- **Uploads:** None — the model is downloaded to you, not the reverse
- **Privacy:** Runs entirely in your browser — nothing is uploaded
- **Cost:** Free, unlimited, no sign-up
- **Detects:** 91 COCO object classes including people and vehicles

## How to use

1. Select your image — the file stays on your device and is never uploaded.
2. Choose the redact.
3. Choose the method.
4. Set the strength.
5. Set the detection confidence.
6. Press Run, then download the result when it is ready.

## FAQ

### Is this good enough for compliance?

Check every result. Object detection misses people who are partly occluded, small in frame, or unusually posed, and a single miss defeats the redaction entirely. Use a low confidence threshold — over-redacting is the safe error — and review the output before publishing.

### Can the blur be reversed?

Not from the exported PNG: the pixels are genuinely destroyed and re-encoded, not overlaid. That is the difference between this and drawing a black rectangle in a PDF or a layered editor, where the original data is still in the file.

### Why does it not detect faces specifically?

The model here detects whole people rather than faces, which for redaction is usually the stronger choice — a person is identifiable from clothing, posture and context, not only their face.

## Sources

- [DETR: End-to-End Object Detection with Transformers](https://arxiv.org/abs/2005.12872) — arXiv

## Related tools

- [CSV Data Anonymiser](https://convertto.tech/t/data-anonymiser): Mask, hash or replace personal data in a CSV so it can be shared safely.
- [Redact PDF](https://convertto.tech/t/redact-pdf): Black out names, figures or images so they are genuinely gone, not just hidden behind a box.
- [EXIF Remover](https://convertto.tech/t/exif-remover): Strip all metadata — including GPS location — from photos before sharing them.
- [PDF Metadata Editor & Remover](https://convertto.tech/t/pdf-metadata-editor): See what a PDF reveals about you, then rewrite it or strip it out entirely.
- [Browser Permissions Inspector](https://convertto.tech/t/permissions-inspector): Check the permission state of camera, microphone, location, notifications and more for this origin.
- [CSS Filter Generator](https://convertto.tech/t/css-filter-generator): Build a CSS filter chain visually and see it applied to a real element as you adjust it.
- [Image Metadata & EXIF Viewer](https://convertto.tech/t/image-metadata-viewer): Inspect EXIF data, camera settings, GPS coordinates and dimensions inside an image.
- [PDF Metadata Viewer](https://convertto.tech/t/pdf-metadata-viewer): Inspect a PDF's title, author, producer, creation date and page dimensions.