BiRefNet one-click startup package, local AI matting.

BiRefNet is a powerful AI image segmentation tool 🖼️ that supports local CPU and GPU operation 🔄. Users can easily use it without a graphics card, enjoying a flexible processing experience ⚙️! The one-click startup package is easy to download with no complicated setup required 🚀.

BiRefNet is an advanced AI matting technology designed for high-resolution binary image segmentation. Presented at the CAAI AIR’24 conference, it excels in image processing and supports local CPU or GPU operation, offering excellent flexibility.

Advantages of Local Deployment: BiRefNet’s highlight is its flexibility, allowing successful operation regardless of whether you have a high-performance GPU.

CPU Deployment: Let Ordinary Computers Play with AI Matting

BiRefNet perfectly supports CPU operation, meaning you can use this powerful AI matting tool even without a dedicated graphics card.

  • No Graphics Card Required: Complete image processing tasks.
  • Sufficient for Personal Use: Although the speed may be slower than GPU, it is perfectly adequate.
  • Home Computer Friendly: Very suitable for ordinary users to use at home.

GPU Acceleration: Experience Lightning Speed

If your computer is equipped with a GPU, BiRefNet can utilize GPU acceleration to improve processing efficiency.

  • CUDA Acceleration: Significantly improve processing speed.
  • Model Invocation: You can call the birefnet-v1-lite model to leverage GPU acceleration for inference.
  • Good CUDA Compatibility: CUDA supports backward compatibility. Even if your CUDA version is 12.6, while the official version supports up to 12.4, Torch can still use CUDA normally.

One-Click Startup Package User Guide

To make it easier for everyone to use, we have created a local one-click startup package. With just a few simple steps, you can use it on your personal computer without worrying about privacy leaks or complex environment configuration issues.

Computer Configuration Requirements


Windows 10/11 64-bit operating system, NVIDIA graphics card with 8G or more of video memory, CUDA >= 12.1

Download and Usage Tutorial

  1. Download the Compressed Package:
    Download address: https://1drv.ms/f/c/D9E7E4EAB666A442/Et_1ta8gbbxJs-FwpBHu2vIBEScnmuVAqE56vQRYfo07Yw

  2. Unzip the File:
    After decompressing, make sure the path does not contain non-English characters. Then, double-click the “run.exe” file to run it.

  3. Browser Access:
    The software will automatically open a browser.