Multimodal Classification on Kinetics-Sounds (test)
72.18Multimodal AccuracyPDMP
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| PDMPBackbone=ResNet-18, Modality=Multimodal (Audio + Visual), Number of frames=3, Visual input resolution=224x224, Audio input size=257x1004, Batch size=64, Epochs=1002026.04 | 72.18 | — | — | 72.67 | |
| MLABackbone=ResNet-18, Modality=Multimodal (Audio + Visual), Number of frames=3, Visual input resolution=224x224, Audio input size=257x1004, Batch size=64, Epochs=1002026.04 | 69.62 | — | — | 69.98 | |
| MIMOt(s)=0.039±0.0152025.11 | 69.6 | 45.07 | 43.13 | — | |
| MMParetoBackbone=ResNet-18, Modality=Multimodal (Audio + Visual), Number of frames=3, Visual input resolution=224x224, Audio input size=257x1004, Batch size=64, Epochs=1002026.04 | 69.13 | — | — | 69.05 | |
| D&RBackbone=ResNet-18, Modality=Multimodal (Audio + Visual), Number of frames=3, Visual input resolution=224x224, Audio input size=257x1004, Batch size=64, Epochs=1002026.04 | 69.1 | — | — | 69.36 | |
| MMParetot(s)=0.689±0.0982025.11 | 68.6 | 48.07 | 37.23 | — | |
| Grad-BlendingBackbone=ResNet-18, Modality=Multimodal (Audio + Visual), Number of frames=3, Visual input resolution=224x224, Audio input size=257x1004, Batch size=64, Epochs=1002026.04 | 67.31 | — | — | 67.68 | |
| OGM-GEBackbone=ResNet-18, Modality=Multimodal (Audio + Visual), Number of frames=3, Visual input resolution=224x224, Audio input size=257x1004, Batch size=64, Epochs=1002026.04 | 66.35 | — | — | 66.76 | |
| AGMBackbone=ResNet-18, Modality=Multimodal (Audio + Visual), Number of frames=3, Visual input resolution=224x224, Audio input size=257x1004, Batch size=64, Epochs=1002026.04 | 65.61 | — | — | 65.99 | |
| PMRBackbone=ResNet-18, Modality=Multimodal (Audio + Visual), Number of frames=3, Visual input resolution=224x224, Audio input size=257x1004, Batch size=64, Epochs=1002026.04 | 65.01 | — | — | 65.13 | |
| ConcatenationBackbone=ResNet-18, Modality=Multimodal (Audio + Visual), Number of frames=3, Visual input resolution=224x224, Audio input size=257x1004, Batch size=64, Epochs=1002026.04 | 64.97 | — | — | 65.21 | |
| OGM-GEt(s)=0.247±0.1192025.11 | 63.73 | 44.1 | 22.57 | — | |
| TCMaxFusion Head=Share Head2026.02 | 63.5 | 43.4 | 43.3 | — | |
| MMPareto2026.02 | 62.7 | 44.8 | 49.4 | — | |
| OGM-GE + OPM2026.02 | 62.5 | 37.6 | 44.1 | — | |
| TCMaxFusion Head=Concat2026.02 | 62.4 | 41 | 41.4 | — | |
| Unimodal Ensemble2026.02 | 62.3 | 43 | 45.8 | — | |
| QMF2026.02 | 62.1 | 44.6 | 44.2 | — | |
| OPM2026.02 | 61.7 | 39.1 | 44.1 | — | |
| MLA2026.02 | 61.1 | 41.6 | 44.9 | — | |
| EWt(s)=0.026±0.0102025.11 | 60.73 | 45.13 | 33.3 | — | |
| MSLt(s)=0.023±0.0072025.11 | 59.83 | 41.2 | 18.67 | — | |
| AGM2026.02 | 57.7 | 36.4 | 29.4 | — | |
| OGM-GE2026.02 | 55.4 | 35.7 | 25.6 | — | |
| Share Head2026.02 | 53.7 | 38.7 | 29.1 | — | |
| Concat2026.02 | 53.5 | 36.8 | 22.2 | — | |
| FiLM2026.02 | 52.6 | — | — | — | |
| Visual-onlyBackbone=ResNet-18, Modality=Visual, Number of frames=3, Visual input resolution=224x224, Batch size=64, Epochs=1002026.04 | 52.36 | — | — | 52.67 | |
| BiGated2026.02 | 49.1 | — | — | — | |
| Audio-onlyBackbone=ResNet-18, Modality=Audio, Audio input size=257x1004, Batch size=64, Epochs=1002026.04 | 48.67 | — | — | 48.89 |