version: '1.0'
module:
  uid: 88f1d52a-123d-42d2-9cee-fc0560dd2589
  name: InsightFace
  version: '1.0'
  description: 'InsightFace Buffalo face feature extraction via ArcFace producing L2-normalized 512-dim embeddings.'
  keywords:
    - Face
    - Recognition
    - Feature
  requirements:
    - seetapsych-face-hub>=0.0.2
    - onnx>=1.20.1
    - onnxruntime>=1.24; sys_platform == 'darwin' and python_version > '3.10'
    - onnxruntime<1.24; sys_platform == 'darwin' and python_version <= '3.10'
    - onnxruntime-gpu>=1.24; sys_platform != 'darwin' and python_version > '3.10'
    - onnxruntime-gpu<1.24; sys_platform != 'darwin' and python_version <= '3.10'
    - opencv-python>=4.11.0.86
    - scikit-image>=0.25.2
packages:
  - uid: a761cb3a-dba8-4b73-bc2b-6a1c02e55a27
    name: FaceFeature-ArcFace(InsightFace)
    description: Extract L2-normalized 512-dim face embeddings for recognition, clustering, or similarity search from 5-point aligned faces
    version: '1.0'
    usage_models: []
    inputs: []
    provides:
      - face/feature
    requires:
      - face/landmarks
    entry:
      method: seetapsych_face_hub.insightface.arcface.load
    models:
      - uid: ee4bd102-8b5f-4b94-a52c-d97aca06a729
        name: w600k_r50.onnx
        version: '1.0'
        recommended: true
        download:
          index: w600k_r50.onnx
          sha256: 4c06341c33c2ca1f86781dab0e829f88ad5b64be9fba56e56bc9ebdefc619e43
          url: https://www.modelscope.cn/models/levalup/cloud-models/resolve/master/insightface/buffalo/w600k_r50.onnx
      - uid: 97f41ab5-297f-435d-add4-98f25aaa6a0d
        name: w600k_mbf.onnx
        version: '1.0'
        download:
          index: w600k_mbf.onnx
          sha256: 9cc6e4a75f0e2bf0b1aed94578f144d15175f357bdc05e815e5c4a02b319eb4f
          url: https://www.modelscope.cn/models/levalup/cloud-models/resolve/master/insightface/buffalo/w600k_mbf.onnx