Skip to content

Latest commit

 

History

History
 
 

arcface_torch

README.md

Arcface Pytorch (Distributed Version of ArcFace)

Contents

Set Up

torch >= 1.6.0

More details see eval.md in docs.

Training

1. Single node, 8 GPUs:

python -m torch.distributed.launch --nproc_per_node=8 --nnodes=1 --node_rank=0 --master_addr="127.0.0.1" --master_port=1234 train.py

2. Multiple nodes, each node 8 GPUs:

Node 0:

python -m torch.distributed.launch --nproc_per_node=8 --nnodes=2 --node_rank=0 --master_addr="ip1" --master_port=1234 train.py

Node 1:

python -m torch.distributed.launch --nproc_per_node=8 --nnodes=2 --node_rank=1 --master_addr="ip1" --master_port=1234 train.py

3.Training resnet2060 with 8 GPUs:

python -m torch.distributed.launch --nproc_per_node=8 --nnodes=1 --node_rank=0 --master_addr="127.0.0.1" --master_port=1234 train.py --network r2060

Speed Benchmark

Image text

ArcFace_torch can train large-scale face recognition training set efficiently and quickly.
When the number of classes in training sets is greater than 300K and the training is sufficient, partial fc sampling strategy will get same accuracy with several times faster training performance and smaller GPU memory.

  1. Training speed of different parallel methods (samples/second), Tesla V100 32GB * 8. (Larger is better)
Method Bs1024-R100-2 Million Identities Bs1024-R50-4 Million Identities Bs512-R50-8 Million Identities
Data Parallel 1 1 1
Model Parallel 1362 1600 482
Fp16 + Model Parallel 2006 2165 767
Fp16 + Partial Fc 0.1 3247 4385 3001
  1. GPU memory cost of different parallel methods (GB per GPU), Tesla V100 32GB * 8. (Smaller is better)
Method Bs1024-R100-2 Million Identities Bs1024-R50-4 Million Identities Bs512-R50-8 Million Identities
Data Parallel OOM OOM OOM
Model Parallel 27.3 30.3 32.1
Fp16 + Model Parallel 20.3 26.6 32.1
Fp16 + Partial Fc 0.1 11.9 10.8 11.1

Evaluation IJBC

More details see eval.md in docs.

Inference

python inference.py --weight ms1mv3_arcface_r50/backbone.pth --network r50

Model Zoo

The models are available for non-commercial research purposes only.

All Model Can be found in here.
Baidu Yun Pan: e8pw
onedrive

MS1MV3

Datasets log backbone IJBC(1e-05) IJBC(1e-04) agedb30 cfp_fp lfw
MS1MV3-Arcface log r18-fp16 92.07 94.66 97.77 97.73 99.77
MS1MV3-Arcface log r34-fp16 94.10 95.90 98.10 98.67 99.80
MS1MV3-Arcface log r50-fp16 94.79 96.46 98.35 98.96 99.83
MS1MV3-Arcface log r100-fp16 95.31 96.81 98.48 99.06 99.85
MS1MV3-Arcface log r2060-fp16 95.34 97.11 98.67 99.24 99.87

Glint360k

Datasets log backbone IJBC(1e-05) IJBC(1e-04) agedb30 cfp_fp lfw
Glint360k-Cosface log r18-fp16-0.1 93.16 95.33 97.72 97.73 99.77
Glint360k-Cosface log r34-fp16-0.1 95.16 96.56 98.33 98.78 99.82
Glint360k-Cosface log r50-fp16-0.1 95.61 96.97 98.38 99.20 99.83
Glint360k-Cosface log r100-fp16-0.1 95.88 97.32 98.48 99.29 99.82

0.1 means sample rate is 0.1.

More details see eval.md in docs.

Test

We test on PyTorch versions 1.6.0, 1.7.1, and 1.8.0. Please create an issue if you are having trouble.

Citation

@inproceedings{deng2019arcface,
  title={Arcface: Additive angular margin loss for deep face recognition},
  author={Deng, Jiankang and Guo, Jia and Xue, Niannan and Zafeiriou, Stefanos},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  pages={4690--4699},
  year={2019}
}
@inproceedings{an2020partical_fc,
  title={Partial FC: Training 10 Million Identities on a Single Machine},
  author={An, Xiang and Zhu, Xuhan and Xiao, Yang and Wu, Lan and Zhang, Ming and Gao, Yuan and Qin, Bin and
  Zhang, Debing and Fu Ying},
  booktitle={Arxiv 2010.05222},
  year={2020}
}