BiRefNet HRSOD (birefnet-hrsod) background removal#
BiRefNet checkpoint for high-resolution salient object detection (HRSOD and DUTS).
Run it: rembg i -m birefnet-hrsod input.jpg output.png setup
birefnet-hrsod results on 34 images
Original, cutout and alpha matte for every Bench v0 image, with its MAE and MGE against the ground truth. Each opens the side-by-side comparison.
People (12)




































Plants (7)





















Objects with Details (6)
Motion (5)
animal (2)
Film Camera (2)
How to run birefnet-hrsod
birefnet-hrsod ships with rembg as the session birefnet-hrsod. The weights download on first use.
Command line
pip install "rembg[cpu,cli]"
rembg i -m birefnet-hrsod input.jpg output.pngPython
from rembg import new_session, remove
from PIL import Image
session = new_session("birefnet-hrsod") # load once, reuse for every image
output = remove(Image.open("input.jpg"), session=session)
output.save("output.png")Other sessions: all rembg models.
BiRefNet HRSOD as a free, open-source remove.bg alternative
remove.bg is a paid cloud service. BiRefNet HRSOD is open weights under MIT, so commercial use is allowed. It runs on your own machine (rembg i -m birefnet-hrsod input.jpg output.png), offline, with no upload and no per-image cost. On Bench v0 it ranks #9 of 17 by accuracy (MAE); withoutBG Open Weights ranks #3.
remove.bg is not in Bench v0, so there is no score against it. See withoutBG Open Weights and remove.bg side by side, or compare all remove.bg alternatives.
Quick take
- Best for: Picking the main subject out of large photos
- Avoid when: Several subjects that should all be kept
- Notes: Salient-object training: it keeps what stands out, which is not always the whole subject.
When a hosted model is worth it
birefnet-hrsod is free and runs on your machine; keep it if the results above are good enough. If you need finer hair and edges, or don't want to run large models locally, rembg also ships the withoutBG Pro Model as the withoutbg session, ranked #1 of 17 by accuracy on Bench v0. Images are sent to withoutBG's servers.
Same command, hosted model
export WITHOUTBG_API_KEY=sk_...
rembg i -m withoutbg input.jpg output.pngSign up for 50 free credits (no credit card) or See its results.












































