InsightFace_Pytorch

Pytorch0.4.1 codes for InsightFace

1. Intro

• This repo is a reimplementation of Arcface(paper), or Insightface(github)

• For models, including the pytorch implementation of the backbone modules of Arcface and MobileFacenet

• Codes for transform MXNET data records in Insightface(github) to Image Datafolders are provided

• Pretrained models are posted, include the MobileFacenet and IR-SE50 in the original paper

2. Pretrained Models & Performance

IR-SE50 @ BaiduNetdisk, IR-SE50 @ Onedrive

Mobilefacenet @ BaiduNetDisk, Mobilefacenet @ OneDrive

3. How to use

• clone

  git clone https://github.com/TropComplique/mtcnn-pytorch.git

3.1 Data Preparation

3.1.1 Prepare Facebank (For testing over camera or video)

Provide the face images your want to detect in the data/face_bank folder, and guarantee it have a structure like following:

data/facebank/
        ---> id1/
            ---> id1_1.jpg
        ---> id2/
            ---> id2_1.jpg
        ---> id3/
            ---> id3_1.jpg
           ---> id3_2.jpg

3.1.2 download the pretrained model to work_space/model

If more than 1 image appears in one folder, an average embedding will be calculated

3.2.3 Prepare Dataset ( For training)

download the refined dataset: (emore recommended)

• emore dataset @ BaiduDrive, emore dataset @ Dropbox

• More Dataset please refer to the original post

Note: If you use the refined MS1M dataset and the cropped VGG2 dataset, please cite the original papers.

• after unzip the files to 'data' path, run :

  python prepare_data.py

  after the execution, you should find following structure:

faces_emore/
            ---> agedb_30
            ---> calfw
            ---> cfp_ff
            --->  cfp_fp
            ---> cfp_fp
            ---> cplfw
            --->imgs
            ---> lfw
            ---> vgg2_fp

3.2 detect over camera:

• 
  

1. download the desired weights to model folder:

• IR-SE50 @ BaiduNetdisk

• IR-SE50 @ Onedrive

• Mobilefacenet @ BaiduNetDisk

• Mobilefacenet @ OneDrive

• 2 to take a picture, run

  python take_pic.py -n name

  press q to take a picture, it will only capture 1 highest possibility face if more than 1 person appear in the camera

• 3 or you can put any preexisting photo into the facebank directory, the file structure is as following:

- facebank/
         name1/
             photo1.jpg
             photo2.jpg
             ...
         name2/
             photo1.jpg
             photo2.jpg
             ...
         .....
    if more than 1 image appears in the directory, average embedding will be calculated

• 4 to start

  python face_verify.py

3.3 detect over video:

```
python infer_on_video.py -f [video file name] -s [save file name]
```

the video file should be inside the data/face_bank folder

• Video Detection Demo @Youtube

3.4 Training:

```
python train.py -b [batch_size] -lr [learning rate] -e [epochs]

# python train.py -net mobilefacenet -b 200 -w 4
```

4. References

• This repo is mainly inspired by deepinsight/insightface and InsightFace_TF

PS

• PRs are welcome, in case that I don't have the resource to train some large models like the 100 and 151 layers model

• Email : treb1en@qq.com