EdgeFace-S for Lumen
EdgeFace (Anjith George, Christophe Ecabert, Hatef Otroshi Shahreza,
Ketan Kotwal, Sébastien Marcel; Idiap Research Institute, 2023), the edgeface_s_gamma_05 model from
Idiap/EdgeFace-S-GAMMA, converted to ONNX for
Lumen's Fast face grouping.
- Input
data: 1×3×112×112, RGB, raw 0–255, a face aligned to ArcFace's 5-point template (normalized to [-1, 1] inside). - Output
fc1: 512 numbers; compare faces by cosine similarity. - Matches the original PyTorch model to cosine 1.00000.
License: CC BY-NC-SA 4.0, as the original weights. Paper: EdgeFace: Efficient Face Recognition Model for Edge Devices, IEEE T-BIOM 2024.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support