Test Results
MuSe 2021 test data phase is closed. Post-challenge evaluation is still possible!
Note that each team gets a total of five chances for evaluation of their method on the test set of each sub-challenge.
All submissions made after 3 August AoE do not count towards the results of the competition. However, they can be used for papers outside the challenge or for the camera-ready version. Please understand that we will reduce the number of scoring iterations after the Challenge. Thanks!
The submission strictly needs to follow the following format (specific examples below!):
The name of the ZIP file containing your predictions should be [sub_challenge]_[team name]_[submission no].zip
The structure should be exactly the same as in the provided folder label_segments of the packages. Zip this directory.
Every row in every test .csv (masked) file should be filled with a valid value corresponding to your prediction (either cont. or a class no.)
The file has to be NaN free and includes all time steps to predict (see masked 'xxx.csv' files)
The format has to be a comma-separated '.csv'.
The name of the prediction files has to correspond exactly to the number.
If only selected targets should be scored, e.g. valence but no arousal, just leave these folders (arousal) empty. However, regardless of the number of targets, this counts as one submission.
You have to be logged in with a Google account in order to see the form above (otherwise it is greyed out).
If you do not have or do not want a google account, you can also submit the results to contact.muse2020@gmail.com. Processing may take longer.
Example: MuSe-Wilder
label_segments [zip this directory with your updated results]
arousal
1.csv
timestamp,segment_id,value
5000,1,-0.3857
5250,1,-0.3857
5500,1,-0.3857
...
2.csv
timestamp,segment_id,value
5000,1,-0.1857
5250,1,-0.1857
5500,1,-0.2857
...
3.csv
... (all test files!)
valence
1.csv
timestamp,segment_id,value
5000,1,0.3857
5250,1,0.3857
5500,1,0.3857
...
2.csv
3.csv
Example: MuSe-Sent
label_segments [zip this folder with your updated results]
arousal
1.csv
start,end,segment_id,class_id
5310,26200,1,0
32640,76810,2,1
76810,127780,3,2
...
2.csv
start,end,segment_id,class_id
6640,42000,1,0
42160,53110,2,1
53120,107310,3,2
...
3.csv
... (all test files!)
valence
1.csv
start,end,segment_id,class_id
5310,26200,1,0
32640,76810,2,1
76810,127780,3,2
...
2.csv
3.csv
... (all test files!)
The same logic corresponds to MuSe-Stress and MuSe-Physio reflecting the structure of your directory label_segments of the package. Fake values are used in the given examples.