Human parsing preprocessing (this directory)
============================================

The code in this directory is third-party code redistributed by NH-Nodes.
It is NOT covered by the MIT license in the repository root.

Provenance
  - OOTDiffusion, preprocess/humanparsing (https://github.com/levihsu/OOTDiffusion)
    License: Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International
    (CC BY-NC-SA 4.0).
  - Reused through FitDiT (https://github.com/BoyuanJiang/FitDiT, CC BY-NC-SA 4.0)
    and ComfyUI-Vton-Mask (https://github.com/karthikg-09/ComfyUI-Vton-Mask), which
    states that it follows FitDiT's licensing terms.
  - Parts originate from Self-Correction-Human-Parsing by Peike Li
    (https://github.com/GoGoDuck912/Self-Correction-Human-Parsing, MIT) and from
    HRNet utilities (Copyright (c) Microsoft, MIT). Their MIT license texts are in
    licenses/ at the repository root.

License for this directory
  CC BY-NC-SA 4.0 — https://creativecommons.org/licenses/by-nc-sa/4.0/legalcode
  You may use and adapt it for NON-COMMERCIAL purposes only, with attribution,
  and adaptations must be shared under the same license.

Changes made by NH-Nodes
  - run_parsing.py: ONNX Runtime sessions are created through ../ort_session.py
    (CUDA request keeps a CPU fallback; clear message when onnxruntime is missing).
  - parsing_api.py: removed the pdb import and sys.path bootstrap.
  - datasets/simple_extractor_dataset.py: removed an unused pdb import.
  Each modified file carries a "Modified by NH-Nodes" header line.

The parsing models (parsing_atr.onnx, parsing_lip.onnx) are downloaded at runtime
and are trained on the ATR and LIP datasets, which are distributed for research use.
See THIRD_PARTY_NOTICES.md in the repository root.
