Now the test_dataset is available: https://simsimpression.unige.ch/datasets/.
You can access it with the same account and password that you use for the training dataset.
The test_dataset contains 10 receivers with face, eye and physiological data but no labels. The given data has the same structure as the training dataset.
Please submit your generated labels for the test data through Codalab directly (Participate->Submit/View Results).
The submission should be in a zip file.
The submission folder structures and names are listed as below:
Your submission folder named 'Label_test' should contain 2 subfolders: warmth and competence.
Under warmth and competence, you have the folders named after the receivers from the test_dataset, e.g. P2002. In each receiver folder, it contains 13 csv files corresponding to each stimulus. The csv files should have the same format (header, sep=’,’) and length as in Training_dataset/Label/competence/PXXXX and Training_dataset/Label/warmth/PXXXX
Please do follow the names, structures and format to have a successful evaluation.
Once your submission finishes running, you can download output from scoring step and check the scores.csv file to check your model performance with each receiver from the test_dataset.
A small reminder:
Jan 3rd is the deadline for abstract submission.
The abstract could be modified after submission, before the final paper submission deadline (14/01/2022)
Detailed paper format and submission could be found here: https://simsimpression.unige.ch/description/