Just a simple face attractiveness ranker
1. Dependencies
- keras2.1.2
- face-recognition1.0.0
2. Dataset
SCUT-FBP5500 is a diverse benchmark database for multi-paradigm facial beauty prediction, which is collected by SCUT. You can download it from their github: https://github.com/HCIILAB/SCUT-FBP5500-Database-Release.
3. Preparing
In this step, a face encoding is extracted from each image and is associated with a label. I use pickle to dump it as training.pkl
4. Training
python3 train.py
Model would be saved as face_rank_model.h5
5. Evaluate
python3 evaluate.py
6. Predict
python3 predict.py