Skip to main content

Command Palette

Search for a command to run...

Local, private background removal without uploads

I wanted a background-removal workflow that did not depend on a cloud service.

Updated
3 min read
 Local, private background removal without uploads
M
Science, technology and engineering. Tech Lead at Bytetravel for Globely. I build open-source code, AI products, MCP servers, agent workflows, Laravel/Inertia applications, browser automation and useful local-first tools.

No account. No upload. No API bill. No monthly limit. Just run it locally and process images on your own machine.

So I built Remove Background Local:

https://github.com/tecnomanu/remove-background-local

Discovery notes

This started as a simple need: I wanted a background-removal tool I could trust with real files, not only demo images. Most existing workflows are either cloud-first, subscription-first or limited by credits.

The useful signal was clear: developers, designers and ecommerce teams often need the same basic operation many times, but they do not always want another SaaS step in the middle of their workflow.

Local processing makes the tool boring in the best way: predictable, private and scriptable.

That is why I am especially interested in feedback around:

  • quality on product photos

  • speed on normal laptops

  • edge cases with hair, plants, glass or semi-transparent objects

  • batch workflows

  • CLI usage inside automation pipelines

If enough people find it useful, the next direction is better presets, easier model selection and more automation hooks.

What it does

  • Drag and drop images in a web UI

  • Paste images from the clipboard

  • Process multiple files in a queue

  • Keep results in persistent local sessions

  • Download PNG, WEBP or JPG

  • Choose transparent or solid-color backgrounds

  • Run from the CLI with rm-bg

  • Open it as a small desktop app with Electron

  • Switch between several segmentation models

  • Use alpha matting for harder edges like hair, plants or semi-transparent details

The default model is ISNet, because it is fast and good enough for most cases. For higher quality, the app also includes BiRefNet models.

Why local-first matters

Background removal often involves product photos, client assets, design drafts or personal images.

For that kind of work, local processing is not only convenient. It is also a better default:

  • private by design

  • no vendor lock-in

  • no rate limits

  • no image upload

  • no hidden cost per batch

Try it

If you already have Node.js and Python 3.9+ installed:

npx -y remove-background-local

Then open:

http://127.0.0.1:7860

For a permanent install:

npm install -g remove-background-local
rm-bg web
rm-bg desktop

Feedback I am looking for

I would especially like feedback from people who work with:

  • ecommerce images

  • design workflows

  • AI image tooling

  • local-first apps

  • privacy-sensitive media workflows

If you try it, I would love to know which model gives you the best quality/speed tradeoff on your machine.

Repo: https://github.com/tecnomanu/remove-background-local

npm: https://www.npmjs.com/package/remove-background-local