Skip to content

Latest commit

 

History

History
70 lines (42 loc) · 1.59 KB

face-recognition-resnet100-arcface.md

File metadata and controls

70 lines (42 loc) · 1.59 KB

face-recognition-resnet100-arcface

The original name of the model is LResNet100E-IR,ArcFace@ms1m-refine-v2.

Use Case and High-Level Description

Deep face recognition net with ResNet100 backbone and Arcface loss

Example

Specification

Metric Value
Type Face recognition
GFLOPs 24.209
MParams 65.131
Source framework MXNet*

Accuracy

Performance

Input

Original Model

Image, name: data, shape: 1,3,112,112, format: B,C,H,W, where:

  • B - batch size
  • C - channel
  • H - height
  • W - width

Channel order is RGB.

Converted Model

Image, name: data, shape: 1,3,112,112, format: B,C,H,W, where:

  • B - batch size
  • C - channel
  • H - height
  • W - width

Channel order is BGR.

Output

Original Model

Face embeddings, name: pre_fc1, shape: 1,512, output data format: B,C, where:

  • B - batch size
  • C - row-vector of 512 floating points values, face embeddings

The net outputs on different images are comparable in cosine distance.

Converted Model

Face embeddings, name: pre_fc1, shape: 1,512, output data format: B,C, where:

  • B - batch size
  • C - row-vector of 512 floating points values, face embeddings

The net outputs on different images are comparable in cosine distance.

Legal Information

LICENSE