Action Recognition on UCF101 (Top-1 and Top-5 Accuracy)
87.4Top-1 AccVITO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| VITOPretraining=Video pretraining, Pretraining Dataset=VideoNet2022.10 | 87.4 | — | |
| BYOLPretraining=Standard image pretraining, Pretraining Dataset=ImageNet2022.10 | 85.6 | — | |
| MOCLRPretraining=Standard image pretraining, Pretraining Dataset=ImageNet2022.10 | 85.5 | — | |
| DINOPretraining=Standard image pretraining, Pretraining Dataset=ImageNet2022.10 | 85.4 | — | |
| L2-RobustPretraining=Robust image pretraining, Pretraining Dataset=ImageNet2022.10 | 83.7 | — | |
| SupervisedPretraining=Supervised, Pretraining Dataset=ImageNet2022.10 | 83.4 | — | |
| Stylized-INPretraining=Robust image pretraining, Pretraining Dataset=SIN+IN2022.10 | 83.3 | — | |
| CycleConPretraining=Video pretraining, Pretraining Dataset=R2V22022.10 | 82.8 | — | |
| SVFormer-stmodel=SVFormer-st, pretrain weight acc=82.92025.12 | 80.2 | — | |
| ReSpikemodel=Hybird (SNN+ANN), pretrain weight acc=73.22025.12 | 77.5 | 93.9 | |
| PredNextSimSiampre-train=ucf101, Initial weights=ImageNet2025.09 | 72.24 | 91.81 | |
| PredNextSimSiampre-train=miniKinetics, Initial weights=ImageNet2025.09 | 71.66 | 92.07 | |
| SimSiampre-train=ucf101, Initial weights=ImageNet2025.09 | 70.32 | 91.56 | |
| SupervisedInitial weights=ImageNet + miniKinetics init2025.09 | 70.02 | 91.62 | |
| STSepmodel=SEW ResNet 18, pretrain weight acc=63.22025.12 | 69.5 | 90.7 | |
| SimSiampre-train=miniKinetics, Initial weights=ImageNet2025.09 | 68.7 | 91.91 | |
| SupervisedInitial weights=ImageNet init2025.09 | 64.42 | 87.36 | |
| VLM-RLAIFLLM Size=7B, Zero-shot=true2024.02 | 62.83 | 85.86 | |
| PredNextBarlowTwinspre-train=miniKinetics2025.09 | 62.75 | 88.66 | |
| PredNextBYOLpre-train=miniKinetics2025.09 | 62.01 | 88.26 | |
| PredNextSimCLRpre-train=miniKinetics2025.09 | 61.06 | 87.21 | |
| PredNextBarlowTwinspre-train=ucf1012025.09 | 59.76 | 84.85 | |
| PredNextSimCLRpre-train=ucf1012025.09 | 59.47 | 85.28 | |
| BYOLpre-train=miniKinetics2025.09 | 59.27 | 86.23 | |
| SimCLRpre-train=miniKinetics2025.09 | 59.03 | 85.96 | |
| PredNextBYOLpre-train=ucf1012025.09 | 58.58 | 83.82 | |
| BarlowTwinspre-train=miniKinetics2025.09 | 58.04 | 85.83 | |
| SimCLRpre-train=ucf1012025.09 | 57.04 | 83.82 | |
| LLaMA-VIDLLM Size=7B, Zero-shot=true, Reproduced=true2024.02 | 56.58 | 82.79 | |
| BYOLpre-train=ucf1012025.09 | 56.41 | 83.18 | |
| BarlowTwinspre-train=ucf1012025.09 | 56.15 | 84.25 | |
| PredNextMoCopre-train=ucf1012025.09 | 54.98 | 82.87 | |
| PredNextSimSiampre-train=ucf1012025.09 | 54.93 | 82.77 | |
| VLM-SFTLLM Size=7B, Zero-shot=true2024.02 | 53.03 | 80.34 | |
| CVRLpre-train=ucf101, Base=SimSiam-based2025.09 | 52.81 | 82.15 | |
| Video-LLaVALLM Size=7B, Zero-shot=true2024.02 | 52.33 | 80.86 | |
| ρSimSiampre-train=ucf101, ρ=12025.09 | 52.05 | 81.75 | |
| PredNextMoCopre-train=miniKinetics2025.09 | 51.6 | 79.65 | |
| Video-ChatGPTLLM Size=7B, Zero-shot=true2024.02 | 51.49 | 79.25 | |
| SimSiampre-train=ucf1012025.09 | 50.81 | 81.07 | |
| PredNextSimSiampre-train=miniKinetics2025.09 | 50.65 | 79.01 | |
| MoCopre-train=ucf1012025.09 | 49.7 | 79.7 | |
| STSepmodel=SEW ResNet 182025.12 | 48.5 | 74.7 | |
| MoCopre-train=miniKinetics2025.09 | 45.63 | 76.55 | |
| SupervisedInitial weights=random init2025.09 | 44.07 | 70.84 | |
| SimSiampre-train=miniKinetics2025.09 | 43.77 | 74.89 | |
| Vanilla+TSNmodel=SEW ResNet 182025.12 | 43.3 | 70.8 | |
| STS ResNetmodel=STS ResNet2025.12 | 42.1 | — |