Multimodal Blended Emotion Recognition on BlEmoRE (val)
44.2ACCpresRank-Aware Selective Fusion
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Rank-Aware Selective FusionEncoder Setting=HiCMAE + 22 video + 13 audio encoders, UDA=true, Top-n=false2026.05 | 44.2 | 22.1 | |
| Rank-Aware Selective FusionEncoder Setting=HiCMAE + 22 video + 13 audio encoders, UDA=true, Top-n=true2026.05 | 43.4 | 21.2 | |
| Rank-Aware Selective FusionEncoder Setting=HiCMAE + 22 video + 13 audio encoders, UDA=false, Top-n=false2026.05 | 40.2 | 22.1 | |
| ImageBind + WavLM2026.05 | 34.5 | 17 | |
| HiCMAE2026.05 | 29.8 | 18 | |
| ImageBind2026.05 | 29 | 13 | |
| Trivial baseline (single emotion)2026.05 | 7.7 | 0 | |
| Trivial baseline (blend)2026.05 | 5.6 | 3.5 |