Face Mask Removal with Region-attentive Face Inpainting

Model Architecture

---

Prerequisites

• python == 3.6.5
• pytorch==1.0.0
• torchvision=0.2.0
• tqdm
• tensorboardX
• scikit-image==any

---

Datasets

We synthesize our own Masked-Faces dataset from the CelebA dataset by incorporating five different types of face masks, including surgical mask, regular mask and scarves, which also cover the neck area. To train a model on the Masked-Faces dataset, download dataset from here.

Model Training

To train the model, you can run:

python train.py

Model Testing

To evaluate the model, you can run:

python test.py