tamminhdiep97 / face_augmentation_tool

a repo to augmentation your face datasets

face_augmentation_tool

A repo to augmentation your face datasets: random cut top half of face-image, random cut bottom-half and put mask on face

Recommend: Use conda to create a virtual env for this tool.

Install dependency Library

1. Create conda env:

   conda create --name tool_env python=3.6
   
   conda activate tool_env

2. Install all dependency by pip

   pip install -r requirements.txt
   pip install torch==1.7.0+cpu torchvision==0.8.1+cpu -f https://download.pytorch.org/whl/torch_stable.html

   see: https://pytorch.org/get-started/previous-versions/ for suitable version of pytorch

3. Download weight file and backbone file of (Resnet50) for Retinaface module:

   • Download 2 file .pth at https://drive.google.com/drive/folders/1_MCIVXaGfsC_ZacBtVyCIX9T21lai75H?usp=sharing
   • Put 2 files above at : ./detector/retinaface/weights/

4. Prepare your data:

   • Your data should contain only face image
   • Your data should following this example path
   • Folder_data
     • folder_1
       • image1.jpg
       • image2.jpg
     • folder_2
       • image3.jpg
       • image4.jpg
     • ...
     • folder_n
       • imagem-1.jpg
       • imagem.jpg
   • This tool will infer through your data folder and execute Augmentation on Prob % of total images in each folder if total images in each folder is bigger than minimum_image in config file.

5. Run scripts:

   python process.py

   For first time run, this tool will download dlib model file for Mask_module, it takes sometime.

Result:

Reference:

• Retinaface: https://github.com/biubug6/Pytorch_Retinaface
• Maskface: https://github.com/aqeelanwar/MaskTheFace