Action Recognition on Something-Something V2 (Acc, Macro-F1)
17.4AccuracySlowFast + PAS
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SlowFast + PASInference Strategy=SlowFast stacked with PAS2025.11 | 17.4 | 15.5 | |
| SlowFast-LLaVAInference Strategy=SlowFast2025.11 | 16.8 | 14.7 | |
| TS-LLaVA + PASInference Strategy=TS-LLaVA stacked with PAS2025.11 | 16.5 | 14.6 | |
| PASInference Strategy=Phase Aggregated Smoothing (Ours)2025.11 | 16.3 | 14.4 | |
| TS-LLaVAInference Strategy=TS-LLaVA2025.11 | 15.1 | 12.6 | |
| Default SettingInference Strategy=Default Setting, Backbone=Qwen2.5-VL-7B-Instruct2025.11 | 14.8 | 13.7 |