Neural networks bundled with Rhino Outline AI
=============================================

The files in the "models" folder are pretrained neural networks in ONNX format. They are data
that Rhino Outline loads with ONNX Runtime; they are not part of the Rhino Outline source code
(GPL v2) and keep their own licences. They are unmodified copies of the files listed below.
The installer from GitHub (.rhi) includes all three networks; the Yak / food4Rhino package
includes u2netp and MobileSAM, and downloads IS-Net the first time it is used.

u2netp.onnx
  Network:  U^2-Net (small), Qin et al., "U^2-Net: Going Deeper with Nested U-Structure for
            Salient Object Detection", Pattern Recognition 2020.
            https://github.com/xuebinqin/U-2-Net
  Licence:  Apache License 2.0 (see Apache-2.0.txt)
  ONNX file from rembg, https://github.com/danielgatis/rembg (MIT, Copyright (c) 2020 Daniel Gatis)
            https://github.com/danielgatis/rembg/releases/download/v0.0.0/u2netp.onnx
  SHA-256:  309c8469258dda742793dce0ebea8e6dd393174f89934733ecc8b14c76f4ddd8

isnet-general-use.onnx
  Network:  IS-Net (DIS), Qin et al., "Highly Accurate Dichotomous Image Segmentation",
            ECCV 2022. https://github.com/xuebinqin/DIS
  Licence:  Apache License 2.0 (see Apache-2.0.txt)
  ONNX file from rembg (MIT, Copyright (c) 2020 Daniel Gatis)
            https://github.com/danielgatis/rembg/releases/download/v0.0.0/isnet-general-use.onnx
  SHA-256:  60920e99c45464f2ba57bee2ad08c919a52bbf852739e96947fbb4358c0d964a

mobile_sam_image_encoder.onnx, sam_mask_decoder_multi.onnx
  Network:  MobileSAM, Zhang et al., "Faster Segment Anything: Towards Lightweight SAM for
            Mobile Applications", 2023. https://github.com/ChaoningZhang/MobileSAM
            The mask decoder is that of Segment Anything (Kirillov et al., Meta AI, 2023),
            https://github.com/facebookresearch/segment-anything
  Licence:  Apache License 2.0 (see Apache-2.0.txt)
  ONNX files from https://huggingface.co/Acly/MobileSAM (MIT)
  SHA-256:  580f5fb648ea1062c0aabc26217aed56921985f03f0cbbd852bba81d760cc749 (encoder)
            8976b90a87ba50a6a72217a5ff994f7d25ce16f2229fcc1ed259e1294c622ffe (decoder)

Downloaded on first use only (not bundled):

BiRefNet-general-bb_swin_v1_tiny-epoch_232.onnx  (AIModel=Best)
  Network:  BiRefNet, Zheng et al., "Bilateral Reference for High-Resolution Dichotomous Image
            Segmentation", CAAI AIR 2024. https://github.com/ZhengPeng7/BiRefNet
  Licence:  MIT
  ONNX file from rembg

ONNX Runtime (onnxruntime.dll, Microsoft.ML.OnnxRuntime.dll) is Copyright (c) Microsoft
Corporation, MIT licence, see onnxruntime-LICENSE.txt and onnxruntime-ThirdPartyNotices.txt.
