Background Remover
Remove the background from any image instantly — free, no account required, and nothing leaves your device. BiRefNet, a high-accuracy segmentation model, runs entirely in your browser and outputs a transparent PNG.
Drop an image here, or browse files
JPG, PNG, WebP — up to 10 MB
How It Works
Upload your image
Drag and drop or click to select a JPG, PNG, or WebP image. Your file is read
directly into browser memory via the FileReader API — nothing is sent anywhere.
Model downloads once
On first use, MODNet (~26 MB) is downloaded and saved to your browser's IndexedDB for offline reuse. A consent dialog appears before any download begins.
AI segments the image
MODNet runs inside a Web Worker via WebAssembly, producing a high-resolution foreground matte. The matte is composited with the original using the Canvas 2D API.
Download transparent PNG
The result PNG preserves full alpha-channel transparency. A before/after view lets you inspect the mask quality before downloading.
Privacy & features
- Your image never leaves your device — no server, no cloud, no tracking.
- Powered by BiRefNet — a state-of-the-art dichotomous image segmentation model.
- One-time model download (~26 MB) saved locally — works offline after first use.
- Outputs a PNG with true alpha transparency — ready for Figma, Canva, PowerPoint, and the web.
- Completely free — no account, no usage limits, no watermarks.
Frequently Asked Questions
Is my image uploaded to your servers?
No. Your image never leaves your device. The entire background removal process runs inside your browser using a Web Worker. Nothing is sent to or stored on any server.
Which image formats are supported?
The tool supports JPG, PNG, and WebP images up to 10 MB.
What format is the download?
The result is a PNG file with full alpha-channel transparency. PNG is the best supported web format for transparent images and is accepted by Figma, Canva, PowerPoint, Photoshop, and most design tools.
Why is the first removal slow?
The first time you use the tool, your browser downloads the MODNet model (~26 MB). A consent dialog with a progress bar explains the download before it begins. After this one-time download, the model is saved to your browser's IndexedDB and subsequent removals start immediately — even offline.
Which AI model powers this tool?
The tool uses MODNet (Xenova/modnet) — a fast, accurate background matting model. It is served as an fp32 ONNX model (~26 MB) and runs entirely via WebAssembly inside a Web Worker. The model license is Apache License 2.0.
Does it work offline?
Yes. After the model is downloaded on your first use, it is saved to your browser's IndexedDB. You can remove backgrounds with no internet connection on every subsequent visit.