face_changing_v2

Final Project of CSCI 1430 with a self-implemented face landmark detection model

1. Install requirements

   pip3 install -r requirements.txt

2. Dataset

   (1) Wider Facial Landmarks in-the-wild (WFLW) is a new proposed face dataset. It contains 10000 faces (7500 for training and 2500 for testing) with 98 fully manual annotated landmarks.

   (2) Download WFLW Training and Testing images at [Google Drive]

   (3) Download WFLW Annotations at Face Annotations

   (4) Unzip above two packages and put them on ./data/WFLW/

   (5) Move Mirror98.txt to WFLW/WFLW_annotations

   (6) Preprocess the dataset

   $ cd data 
   $ python3 SetPreparation.py

3. Train and Test

   (1) Train:

   $ python3 train.py

   (2) Use tensorbnoard to view the loss changing:

   $ tensorboard  --logdir=./checkpoint/tensorboard/

   (3) Test:

   $ python3 test.py

4. Face Changing

   (1) Extract and save the video frames

   $ python3 main.py

   (2) Create the output video using saved frames

   $ python3 transfer_to_video.py

5. Result

6. Report and Poster

   Final report: https://drive.google.com/file/d/1hst6AY2cbunJz6_fLI4KINTOAoNz8cAf/view?usp=sharing

   Poster: https://drive.google.com/file/d/1fDGP9QRNnN8bLNO9HHJ1RIW31amfG5Jz/view?usp=sharing

7. Reference

   PFLD: A Practical Facial Landmark Detector https://arxiv.org/pdf/1902.10859.pdf