Natural Language Queries on Ego4D NLQ v2 (val)
31.63R@1 (IoU=0.3)OSGNet
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| OSGNetFeature=E+I2025.05 | 31.63 | 22.03 | 57.91 | 45.19 | |
| GroundVQAFeature=E+I, Published=CVPR 2024, NaQ pretraining=true, reproduced=true2025.05 | 29.68 | 20.23 | 52.17 | 37.83 | |
| EgoVideoFeature=EgoVideo, Published=CVPRW 2024, NaQ pretraining=true2025.05 | 28.65 | 19.73 | 53.3 | 40.42 | |
| GroundNLQFeature=E+I, Published=CVPRW 2023, NaQ pretraining=true2025.05 | 26.98 | 18.83 | 53.56 | 40 | |
| Hand Trajectory FusionModel=traj. fusion2026.06 | 26.54 | 18.5 | 52.37 | — | |
| GroundNLQModel=baseline2026.06 | 25.77 | 17.11 | 51.87 | — | |
| EgoEnvFeature=E+I+C+EgoEnv, Published=NeurIPS 2024, NaQ pretraining=true2025.05 | 25.37 | 15.33 | — | — | |
| NaQFeature=E+I+C, Published=CVPR 2023, NaQ pretraining=true, reproduced=true2025.05 | 24.1 | 15.03 | 29.9 | 20.85 | |
| ASLFeature=E+I, Published=CVPRW 2023, NaQ pretraining=true2025.05 | 22.62 | 15.64 | 46.86 | 32.16 |