BiRefNet
High-resolution dichotomous segmentation — state-of-the-art background removal.
- Stars
- 4.3K
- Forks
- 338
- Open issues
- 33
- Last commit
- 2026-10-11
Live from the GitHub API · updated 2026-10-15
What is BiRefNet?
BiRefNet is a segmentation model that produces very clean cut-outs at high resolution, including hair and thin structures. It is available on Hugging Face and integrated into rembg and ComfyUI nodes.
Key features
- High-resolution, detailed foreground masks
- Pretrained weights for general, portrait and matting use
- Hugging Face transformers loading
System requirements
- Platforms
- Windows · macOS · Linux
- GPU
- NVIDIA recommended
- Python
- 3.9+
How to install BiRefNet
Summarized from the project’s README. Projects change quickly — check the GitHub page for the latest instructions.
Python
- 1Clone and install.
git clone https://github.com/ZhengPeng7/BiRefNet.git cd BiRefNet pip install -r requirements.txt - 2Or load it directly from Hugging Face in your own code.
from transformers import AutoModelForImageSegmentation model = AutoModelForImageSegmentation.from_pretrained("ZhengPeng7/BiRefNet", trust_remote_code=True)
Strengths
- Best-in-class edge quality
- Permissive license
Limitations
- Research code; easiest via rembg or ComfyUI
BiRefNet alternatives compared
| Software | Stars | License | Platforms | VRAM |
|---|---|---|---|---|
| BiRefNet High-resolution dichotomous segmentation — state-of-the-art background removal. | 4.3K | MIT | — | |
| rembg Remove image backgrounds from the command line, Python or an HTTP server. | 25.1K | MIT | — | |
| Segment Anything (SAM) Meta’s promptable segmentation model — click to cut out any object. | 55K | Apache-2.0 | — | |
| Upscayl Free desktop app that upscales images with AI. | 50.6K | AGPL-3.0 | — | |
| Real-ESRGAN Practical image and video super-resolution models. | 37.1K | BSD-3-Clause | — | |
| GFPGAN Blind face restoration for old and low-quality photos. | 37.7K | Apache-2.0 | — |
FAQ
Is BiRefNet free?
Yes — BiRefNet is open-source software released under the MIT license. Check the license (and any model weight licenses) before commercial use.
Does BiRefNet run on Windows, Mac and Linux?
BiRefNet supports: Windows, macOS, Linux.
What hardware do I need for BiRefNet?
NVIDIA recommended
How do I install BiRefNet?
1. Clone and install. 2. Or load it directly from Hugging Face in your own code.
What are the best BiRefNet alternatives?
Popular alternatives include rembg, Segment Anything (SAM), Upscayl, Real-ESRGAN. See the comparison table on this page.
Can I use BiRefNet online without installing it?
If you just want results, AiLoft runs top image models such as GPT Image 2.5 and Nano Banana Pro in the browser — no GPU or setup required.
Skip the install — generate images in your browser
No GPU, drivers or Python needed. Use GPT Image 2.5 and Nano Banana Pro online and compare them side by side.