Error in Emotion_ferplus output labels about models HOT 3 CLOSED

@anirudhacharya for FER+ we relabeled the Kaggle dataset with more epople from crowd source and added Contempt. The new label is in https://github.com/Microsoft/FERPlus

from models.

@ebarsoum So I would be right in taking these as actual labels -
{'neutral':0, 'happiness':1, 'surprise':2, 'sadness':3, 'anger':4, 'disgust':5, 'fear':6, 'contempt':7}

And taking the output as activations which need to pass through a softmax function to get the probability distribution of the different emotions?

from models.

@anirudhacharya Yes, you will need to route the output of the model through a softmax function to get the probabilities across the 8 classes (this ferplus model uses the above 8 classes and not the original 7)

from models.