Evaluation
Metrics
Metrics
The submissions will be evaluated based on the ROC-AUC score. In other words, the score is the average of the individual AUCs of each predicted emotion.
The description of the ROC-AUC score can be found here: https://en.wikipedia.org/wiki/Receiver_operating_characteristic#Area_under_the_curve
Input & Output formats
Input & Output formats
Download an example from our CodaLab. The data archive contains a folder with 2 input training image + text transcription file examples (provider by organizers) and an output text file example (submitted by participants).