Action Spotting on SoccerNet v2 (challenge)
68.38Average-mAP (Tight 1-5s)COMEDIAN (ViViT-T - ens.)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| COMEDIAN (ViViT-T - ens.)Input=End-to-end with images (I), Backbone=ViViT-T, Ensemble=3 seeds, Context=Challenge 2023 submission2023.09 | 68.38 | — | — | |
| Spivak*Input=Feature extractor (F), Context=Challenge 2023 submission2023.09 | 68.33 | — | — | |
| Soares et al.Competition Context=2022 challenge; 1st, trained on train+val+test=true2022.07 | 67.81 | 72.84 | 60.17 | |
| SpivakInput=Feature extractor (F), Context=Challenge 2022 leaderboard2023.09 | 67.81 | — | — | |
| E2E-SpotBackbone=800MF, Competition Context=2022 challenge; 2nd, trained on train+val+test=true2022.07 | 66.73 | 74.84 | 53.21 | |
| E2E-SpotInput=End-to-end with images (I), Context=Challenge 2022 leaderboard2023.09 | 66.73 | — | — | |
| E2E-SpotBackbone=800MF2022.07 | 66.01 | 72.76 | 51.65 | |
| Faster-TADInput=Feature extractor (F), Context=Challenge 2022 leaderboard2023.09 | 64.88 | — | — | |
| E2E-SpotBackbone=200MF2022.07 | 63.28 | 70.41 | 45.98 | |
| Transformer-ASInput=End-to-end with images (I), Context=Challenge 2022 leaderboard2023.09 | 52.04 | — | — | |
| Zhou et al.Competition Context=2021 challenge; 1st2022.07 | 49.56 | 54.42 | 45.42 | |
| BaiduInput=Feature extractor (F), Context=Challenge 2022 leaderboard2023.09 | 49.56 | — | — | |
| NetVLAD++2022.07 | 43.99 | — | — | |
| RMS-Net2022.07 | 27.69 | — | — |