InsightFace-tensorflow

This is a tensorflow implementation of paper "ArcFace: Additive Angular Margin Loss for Deep Face Recognition". This implementation aims at making both usage of pretrained model and training of your own model easier. Whether you just want to use pretrained model to do face recognition/verification or you want train/finetune your own model, this project can give you a favor. An introduction on face recognition losses can be found here(in Chinese).

The implementation referred to the official implementation in mxnet and the previous third-party implementation in tensorflow.

• InsightFace-tensorflow
  • TODO List
  • Running Environment
  • Usage of Pretrained Model
    • Pretrained Model
    • Model Evaluation
    • Extract Embedding with Pretrained Model
  • Train Your Own Model
    • Data Prepare
    • Train with Softmax
    • Finetune with Softmax

TODO List

1. Train with softmax [done!]
2. Model evaluation [done!]
3. Finetune with softmax [done!]
4. Get embedding with pretrained model [done!]
5. Train with triplet loss [todo]
6. Finetune with triplet loss [todo]
7. Backbones
   7.1 ResNet [done!]
   7.2 ResNeXt [todo]
   7.3 DenseNet [todo]
8. Losses
   8.1 Arcface loss [done!]
   8.2 Cosface loss [todo]
   8.3 Sphereface loss [todo]
   8.4 Triplet loss [todo]
9. Face detection and alignment [todo]

Running Environment

• python 3.6
• scipy, numpy (Anaconda 3 recommended)
• tensorflow 1.7.0
• mxnet 1.3.1 (only needed when reading mxrec file)

Usage of Pretrained Model

Here we open our pretrained models for easier application of face recognition or verification. Codes on model evaluation and extracting embedding from face images are supplied.

Pretrained Model

Pretrained models and their accuracies on validation datasets are shown as following:

config	lfw	calfw	cplfw	agedb_30	cfp_ff	cfp_fp	vgg2_fp	steps	download
ms1m_100	99.53%	93.92%	87.85%	94.18%	99.29%	94.73%	93.9%	334k	baidu
ms1m_100	99.53%	94.68%	89.75%	95.20%	99.54%	96.30%	94.84%	1006k	baidu, google
ms1m_200	99.43%	94.40%	88.23%	94.58%	99.29%	94.77%	93.9%	200k	baidu

Model Evaluation

You can evaluate a pretrained model with evaluate.py by specifying the config path and model path, for example:

python evaluate.py 
--config_path=./configs/config_ms1m_100.yaml 
--model_path=$DIRECTORY_TO_PRETRAINED_MODEL$/best-m-150000

This will evaluate the pretrained model on validation datasets specified in the config file. If you want to evaluate the model on other validation dataset, you can specify it by --val_data as following:

python evaluate.py 
--config_path=./configs/config_ms1m_100.yaml 
--model_path=$DIRECTORY_TO_PRETRAINED_MODEL$/best-m-150000 
--val_data=$DIRECTORY_TO_VAL_DATA$/xxx.bin

Extract Embedding with Pretrained Model

You can extract embedding from face images with get_embd.py by the following script:

python get_embd.py 
--config_path=./configs/config_ms1m_100.yaml 
--model_path=$DIRECTORY_TO_PRETRAINED_MODEL$/best-m-150000 
--read_path=$PATH_TO_FACE_IMAGES$
--save_path=$SAVING_DIRECTORY$/embd.pkl

where config_path and model_path specify the config file and pretrained model respectively. read_path is path to face images, that can be a path to one image or a directory with only images in it. save_path specifies where to save the embedding. The saved file is a dict with image file name as key, the corresponding embedding as value, and can be loaded with pickle in python. Note that face images should be well cropped here.

Train Your Own Model

If you want train your own model from scratch, or finetune pretrained model with your own data, here is what you should do.

Data Prepare

The official InsightFace project open their training data in the DataZoo. This data is in mxrec format, you can transform it to tfrecord format with ./data/generateTFRecord.py by the following script:

python generateTFRecord.py 
--mode=mxrec
--image_size=112
--read_dir=$DIRECTORY_TO_THE_TRAINING_DATA$
--save_path=$DIRECTORY_TO_SAVE_TFRECORD_FILE$/xxx.tfrecord

Or, if you want to train the model with your own data, you can prepare the tfrecord file by the following script:

python generateTFRecord.py 
--mode=folders
--image_size=112
--read_dir=$DIRECTORY_TO_THE_TRAINING_DATA$
--save_path=$DIRECTORY_TO_SAVE_TFRECORD_FILE$/xxx.tfrecord

Here, the read_dir should be the directory to your own face images, where images to one person are saved in one folder. The directory should have a structure like this:

read_dir/
  - id1/
    -- id1_1.jpg
    ...
  - id2/
    -- id2_1.jpg
    ...
  - id3/
    -- id3_1.jpg
    -- id3_2.jpg
    ...
  ...

Train with Softmax

To train your own model with softmax, firstly you should prepare a config file like those in ./configs. It is recommended to modify one example config file to your own config. Secondly, the following script starts training:

python train_softmax.py --config_path=./configs/config_ms1m_100.yaml

Finetune with Softmax

To finetune a pretrained model with your own data, you should prepare a finetune config file like ./configs/config_finetune.yaml, and start training by the following script:

python finetune_softmax.py --config_path=./configs/config_finetune.yaml