Zeroshot-visual-emotion-recognition

Official implementation of the IntelliSys2022 paper: Zero-Shot Visual Emotion Recognition by Exploiting BERT

Youtube: https://www.youtube.com/watch?v=jtQv9ICLmFM

Project descriptions

The model is trained with 6 basic emotions(anger, disgust, fear, joy, sadness, surprise)
Through zero-shot learning, the model is capable of predicting unseen emotion classes not included in the training set, which are amusement, awe, excitement, and contentment

Dataset Download links

glove.6B.zip: https://nlp.stanford.edu/projects/glove/
emotion6: http://chenlab.ece.cornell.edu/downloads.html
artphoto: https://www.imageemotion.org/

How to train

python train.py --glove_path ./glove.6B/glove.6B.300d.txt \ 
                    --glove_vec glove_vec.pkl \
                    --emotion6 C:\data\Emotion6\images \
                    --emotion_glove_vec emotion_glove_vec.pkl

Predict unseen emotions

python test.py --dataset artphoto --source C:/data/artphoto