swayhrl / abaw

Emotion Recognition for In-the-wild Videos

Code for the seven basic expression classification track of Affective Behavior Analysis in-the-wild (ABAW) Competition held in conjunction with the IEEE International Conference on Automatic Face and Gesture Recognition (FG) 2020.

Requirements

• Python3
• PyTorch
• TorchVision
• NumPy
• Aff-Wild2 Dataset
• Pre-trained models (optional)

Downloads:

Modal files: link

Dataset: link

Training

Put train.py, face_imgs (from downloads, or use official cropped_aligned), pre-trained modal (optional), EXPR_Set (official annotation) under the same path.

python3 train.py

Testing

Put test.py, face_imgs (from downloads, or use official cropped_aligned), modal, expression_test_set.txt, frames.txt under the same path.

python3 test.py