Action Segmentation on COIN (test)
72.8Frame AccuracyUnLoc-L
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| UnLoc-LModel scale=Large, Prompting strategy=prompt ensembling2023.08 | 72.8 | — | 47.7 | |
| UnLoc-LModel scale=Large, Prompting strategy=1st prompt2023.08 | 72.6 | — | 47 | |
| UniVL2023.08 | 70 | — | — | |
| VideoCLIPEvaluation Protocol=Fine-tuned2021.09 | 68.7 | — | — | |
| VideoCLIP2023.08 | 68.7 | — | — | |
| TACO2023.08 | 68.4 | — | — | |
| VLM2023.08 | 68.4 | — | — | |
| UnLoc-BModel scale=Base, Prompting strategy=1st prompt2023.08 | 68 | — | 36.2 | |
| MIL-NCEEvaluation Protocol=Supervised2021.09 | 61 | — | — | |
| MIL-NCE2023.08 | 61 | — | — | |
| VideoCLIPEvaluation Protocol=Zero-shot2021.09 | 58.9 | — | — | |
| BaselineStrategy=predict all background2023.08 | 58.9 | — | 0 | |
| ActBERTEvaluation Protocol=Supervised2021.09 | 57 | — | — | |
| ActBERT2023.08 | 57 | — | — | |
| ActBERTregion cues=true, Evaluation Protocol=fine-tune2020.11 | 56.95 | — | — | |
| CBTEvaluation Protocol=Supervised2021.09 | 53.9 | — | — | |
| ActBERTregion cues=false, Evaluation Protocol=fine-tune2020.11 | 52.1 | — | — | |
| TCFPN-ISBA2020.11 | 34.3 | — | — | |
| TCFPN-ISBAEvaluation Protocol=Supervised2021.09 | 34.3 | — | — | |
| VGGEvaluation Protocol=Supervised2021.09 | 25.8 | — | — | |
| VGG2020.11 | 25.79 | — | — | |
| NN-ViterbiEvaluation Protocol=Supervised2021.09 | 21.2 | — | — | |
| NN-Viterbi2020.11 | 21.17 | — | — | |
| CBTNet=S3D, Pretraining Dataset=Kinetics-600 + HowTo100M, Pretraining Labels=true2019.12 | — | 53.9 | — | |
| I3D (Supervised)Net=I3D, Pretraining Dataset=Kinetics-400, Pretraining Labels=true2019.12 | — | 52.9 | — | |
| I3D (Supervised)Net=I3D, Pretraining Dataset=Kinetics-700, Pretraining Labels=true2019.12 | — | 54.2 | — | |
| MIL-NCENet=I3D, Pretraining Dataset=HowTo100M, Pretraining Labels=false2019.12 | — | 59.4 | — | |
| MIL-NCENet=S3D, Pretraining Dataset=HowTo100M, Pretraining Labels=false2019.12 | — | 61 | — | |
| ResNet-50Net=R50, Pretraining Dataset=ImageNet, Pretraining Labels=true2019.12 | — | 52 | — |