Egocentric Human Activity Recognition on EgoExo4D
84.92Accuracy @1COMODO
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| COMODOSelf-supervised=true2025.03 | 84.92 | 98.28 | 99.59 | |
| MantisSelf-supervised=false2025.03 | 84.22 | 97.95 | 99.41 | |
| COMODOTrain=Ego4D, Supervised fine-tuning on source=false2025.03 | 82.54 | 98.02 | 99.56 | |
| AttendSelf-supervised=false2025.03 | 81.45 | 98.2 | 99.66 | |
| MantisTrain=Ego4D, Supervised fine-tuning on source=true2025.03 | 80.79 | 97.82 | 99.53 | |
| IMU2CLIPTrain=Ego4D, Supervised fine-tuning on source=false2025.03 | 79.7 | 97.68 | 99.45 | |
| IMU2CLIPSelf-supervised=true2025.03 | 79.46 | 97.68 | 99.44 | |
| MOMENT-smallSelf-supervised=false2025.03 | 79.26 | 97.04 | 99.33 | |
| CrossHARSelf-supervised=false2025.03 | 77.14 | 96.95 | 99.31 | |
| DeepConvLSTMSelf-supervised=false2025.03 | 76.51 | 96.89 | 99.37 | |
| MantisSelf-supervised=true2025.03 | 76.47 | 96.98 | 99.21 | |
| IMUGPT 2.0 (V+R)Self-supervised=false, Training Data Type=virtual + real IMU data2025.03 | 74.93 | 96.45 | 99.13 | |
| InformerSelf-supervised=false2025.03 | 74.6 | 96.28 | 99.17 | |
| TimesNetSelf-supervised=false2025.03 | 69.2 | 93.44 | 98.04 | |
| CrossHARTrain=Ego4D, Supervised fine-tuning on source=false2025.03 | 68.22 | 95.11 | 98.64 | |
| MOMENT-smallSelf-supervised=true2025.03 | 68.14 | 93.55 | 98.43 | |
| CrossHARSelf-supervised=true2025.03 | 67.8 | 94.67 | 98.66 | |
| IMUGPT 2.0 (V)Self-supervised=false, Training Data Type=virtual-only IMU data2025.03 | 44.76 | 56.8 | 73.93 | |
| DLinearSelf-supervised=false2025.03 | 32.2 | 59.91 | 78.16 |