Browser background removal — third-party notices (2026-10-09)

U2NETP model: U^2-Net by Xuebin Qin, Zichen Zhang, Chenyang Huang,
Masood Dehghan, Osmar R. Zaiane and Martin Jagersand.
Paper: U^2-Net: Going Deeper with Nested U-Structure for Salient Object Detection.
Project: https://github.com/xuebinqin/U-2-Net
Code and released models: Apache License 2.0; see /licenses/u2net-LICENSE.txt.
Author's model-license statement and training-data reminder:
https://github.com/xuebinqin/U-2-Net/issues/208?timeline=1
U2NETP was trained on DUTS. Training-dataset rights are not controlled by
the model authors. The model license does not clear all training-data rights.
Dataset: https://saliencydetection.net/duts/

ONNX conversion distributed by rembg:
https://github.com/danielgatis/rembg/releases/download/v0.0.0/u2netp.onnx
Unmodified artifact: 4,574,861 bytes.
Upstream expected MD5: 8e83ca70e441ab06c318d82300c84806
Project-measured SHA-256:
309c8469258dda742793dce0ebea8e6dd393174f89934733ecc8b14c76f4ddd8
This application uses its own browser preprocessing code; rembg Python
code is not bundled. rembg's MIT code license does not replace model terms.

ONNX Runtime Web 1.30.0 by Microsoft Corporation: MIT License.
Project: https://github.com/microsoft/onnxruntime
License: /licenses/onnxruntime-LICENSE.txt
Runtime dependencies: /licenses/onnxruntime-ThirdPartyNotices.txt
The runtime files are copied unmodified from the pinned npm package.

Application browser worker and UI were authored for Image Studio.
Keep these notices and upstream licenses when redistributing the model
and runtime. This local preview does not make a guarantee of removal
quality or that all commercial image and dataset rights have been cleared.