Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

MOSEI

Benchmarks

Task NameDataset NameSOTA ResultTrend
Multimodal Sentiment AnalysisMOSEI
MAE0.496
168
Multimodal Sentiment AnalysisMOSEI (test)
MAE0.523
49
Emotion RecognitionMOSEI
Accuracy (7-Class)51.88
26
Binary Sentiment ClassificationMOSEI (test)
Accuracy82.5
16
Multimodal Sentiment AnalysisMOSEI 2018b (test)
Acc2 (pos/neg)85.88
13
Sentiment AnalysisMOSEI (held-out)
F1 Score72.4
8
Sentiment Analysis (SEN)MOSEI S
Binary Weighted F177.5
7
Emotion Recognition (EMO)MOSEI E
Mean Weighted Accuracy61.4
7
Streaming Social Task DetectionMOSEI
Accuracy85.64
7
Sentiment ClassificationMOSEI (test)
Accuracy (2 Class)85
7
Multi-label ClassificationMOSEI
F1 (Happy)72.7
5
Binary ClassificationMOSEI
F1 (Happy)71.7
5
Cross-modal retrieval (Video)MOSEI
R@139.2
4
Cross-modal retrieval (Language)MOSEI
Recall@134.6
4
Cross-modal retrieval (Audio)MOSEI
R@134.6
4
Sentiment AnalysisMOSEI
Number of Parameters256,453
4
Audiovisual Sentiment AnalysisMOSEI
Accuracy (AV)67.19
3
Model SelectionMOSEI (unseen)
Performance99.35
1
Showing 18 of 18 rows