bg-remove: Remove Image Backgrounds in Your Browser

Overview

bg-remove is a React and Vite web app for removing image backgrounds on the client side. It creates an alpha mask with a machine-learning model, then lets you keep the image transparent or apply a color or image background.

Because processing happens in the browser, it is useful for people who want a privacy-focused alternative to uploading images to a background-removal service. It supports a default model for broad browser compatibility and an optional WebGPU-based alternative.

Key Features

• Remove image backgrounds with a one-click workflow.
• Process images locally in the browser, without server uploads.
• Use RMBG-1.4 by default, with WebAssembly processing.
• Choose MODNet when WebGPU is available.
• Set a custom background color or image, or retain transparency.
• Download the result with a transparent or colored background.
• Use IndexedDB through Dexie.js for local file management.

Use Cases

• Individuals preparing transparent product images for listings without sending files to a remote service.
• Designers making quick image cutouts for mockups or other visual assets.
• Developers evaluating browser-based image processing with Transformers.js.
• Privacy-conscious users who want to remove backgrounds from personal images locally.

Project Facts

• Language: TypeScript
• License: MIT
• Stars: 1k
• Forks: 119
• Topics: background-image, background-image-remover, background-remover, images
• Archived: No

Getting Started

git clone https://github.com/addyosmani/bg-remove.git
cd bg-remove && npm install && npm run dev

See the README for setup and usage details. Try the app at bg.addy.ie.

Alternatives

• photopea: Photopea is a full-featured browser image editor with manual background-removal tools, while bg-remove focuses on local machine-learning removal.

Considerations

• Processing and model inference take place in the browser, so available device resources and browser performance can affect the experience.
• WebGPU acceleration is optional and requires a compatible browser; the README describes its availability as limited to browsers with WebGPU support, such as Chrome Canary with WebGPU flags enabled.
• The default RMBG-1.4 path is intended for modern browsers and uses WebAssembly; MODNet is an alternative rather than the default.