Vague Query-based Affective Video Understanding on VQAU
19.55mIoUAffectSeek
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| AffectSeekModel Category=Multi-Agents2026.05 | 19.55 | 33.33 | 14.44 | 3.33 | 15.91 | 4.55 | 1.14 | 7.09 | |
| Emotion-LLaMaModel Category=Fine-tuned Models2026.05 | 12.95 | 8.57 | 3.29 | 1.99 | 3.79 | 1.55 | 0.11 | 4.52 | |
| Gemini3.1-ProModel Category=Commercial Models2026.05 | 10.61 | 14.88 | 6.81 | 2.63 | 8.14 | 3.72 | 1.48 | 5.12 | |
| Qwen3.5Model Category=Commercial Models2026.05 | 9.02 | 12.9 | 5.97 | 2.16 | 9.6 | 4.39 | 1.52 | 4.9 | |
| Emotion-LLaMa-v2Model Category=Fine-tuned Models2026.05 | 8.91 | 13.79 | 5.62 | 2.4 | 7.1 | 2.76 | 1.28 | 3.25 | |
| Kimi-2.5Model Category=Commercial Models2026.05 | 6.68 | 7.84 | 3.94 | 1.62 | 3.93 | 1.87 | 0.77 | 3.02 | |
| GPT-5.4Model Category=Commercial Models2026.05 | 6.26 | 8.22 | 3.88 | 1.54 | 4.62 | 2.14 | 0.86 | 5.15 | |
| AffectGPTModel Category=Fine-tuned Models2026.05 | 5.22 | 7.85 | 3.59 | 1.06 | 3.96 | 1.89 | 0.67 | 3.5 |