Temporal Action Localization on ActivityNet 1.3 (50%-50%)
32.5Avg mAPConTrans
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| ConTransData split Train:Test=50:50, Protocol=Zero-shot, Setting=Open-set2026.05 | 32.5 | 45.3 | 30.3 | 10.5 | |
| STOV-TALTraining=supervised, VAE=false2026.01 | 29.6 | — | — | — | |
| PyraTokTraining=zero-shot, VAE=true2026.01 | 29.11 | — | — | — | |
| STOV-TALTraining=zero-shot, VAE=false2026.01 | 28 | — | — | — | |
| GAPData split Train:Test=50:50, Protocol=Zero-shot, Setting=Open-set2026.05 | 26.4 | 41.6 | 26.2 | 6.1 | |
| GRIZALData split Train:Test=50:50, Protocol=Zero-shot, Setting=Open-set2026.05 | 25.7 | 39.9 | 25.7 | 6.6 | |
| mProTEAData split Train:Test=50:50, Protocol=Zero-shot, Setting=Open-set2026.05 | 25.6 | 41.8 | 24.6 | 6.1 | |
| LARPTraining=zero-shot, VAE=true2026.01 | 25.53 | — | — | — | |
| ZEETADTraining-based=Yes2026.05 | 24.9 | 39.2 | 25.7 | 3.1 | |
| SweetTokTraining=zero-shot, VAE=true2026.01 | 24.53 | — | — | — | |
| OmniTokenizerTraining=zero-shot, VAE=true2026.01 | 22.48 | — | — | — | |
| DeTALTraining=supervised, VAE=false2026.01 | 22.4 | — | — | — | |
| VideoVae+Training=zero-shot, VAE=true2026.01 | 21.37 | — | — | — | |
| STALETraining=supervised, VAE=false2026.01 | 20.5 | — | — | — | |
| STALETraining Protocol=Training-based2024.04 | 20.5 | 32.1 | 20.7 | 5.9 | |
| STALETraining-based=Yes2026.05 | 20.5 | 32.1 | 20.7 | 5.9 | |
| STALEData split Train:Test=50:50, Protocol=Zero-shot, Setting=Open-set2026.05 | 20.5 | 32.1 | 20.7 | 5.9 | |
| EffPromptTraining Protocol=Training-based2024.04 | 19.6 | 32 | 19.3 | 2.9 | |
| EffPromptTraining-based=Yes2026.05 | 19.6 | 32 | 19.3 | 2.9 | |
| EffPromptData split Train:Test=50:50, Protocol=Zero-shot, Setting=Open-set2026.05 | 19.6 | 32 | 19.3 | 2.9 | |
| TEGUTraining-based=No2026.05 | 18.5 | 33.5 | 19.3 | 2.6 | |
| CLIP16 w/ DetectorBackbone=CLIP16, Detector Usage=true, Training Protocol=Training-based2024.04 | 16 | 28 | 16.4 | 1.2 | |
| Detector + CLIPTraining-based=Yes2026.05 | 16 | 28 | 16.4 | 1.2 | |
| B-IData split Train:Test=50:50, Protocol=Zero-shot, Setting=Open-set2026.05 | 16 | 28 | 16.4 | 1.2 | |
| T3ALTraining Protocol=Zero-shot2024.04 | 14.3 | 25.8 | 13.9 | 3.1 | |
| T3ALTraining-based=No2026.05 | 14.3 | 25.8 | 13.9 | 3.1 | |
| T3ALT=0Adaptation Steps (T)=0, Training Protocol=Zero-shot2024.04 | 13.3 | 24.2 | 13 | 2.8 | |
| B-IIData split Train:Test=50:50, Protocol=Zero-shot, Setting=Open-set2026.05 | 12.9 | 32.1 | 20.7 | 3.7 | |
| CoCaTraining Protocol=Zero-shot2024.04 | 1.1 | 2.3 | 1 | 0.2 | |
| CLIP16Backbone=CLIP16, Training Protocol=Zero-shot2024.04 | 0.3 | 0.8 | 0.3 | 0 | |
| CLIP32Backbone=CLIP32, Training Protocol=Zero-shot2024.04 | 0.2 | 0.4 | 0.2 | 0 | |
| UnLocTraining Protocol=Training-based2024.04 | — | 43.7 | — | — | |
| UnLocTraining-based=Yes2026.05 | — | 43.7 | — | — | |
| UnLocData split Train:Test=50:50, Protocol=Zero-shot, Setting=Open-set2026.05 | — | 43.7 | — | — |