Vision And Language Navigation On Vln
评估指标
error
length
oracle success
spl
success
评测结果
各个模型在此基准测试上的表现结果
模型名称 | error | length | oracle success | spl | success | Paper Title | Repository |
|---|---|---|---|---|---|---|---|
| Global Normalization | 2.99 | 686.86 | 0.99 | 0.01 | 0.74 | - | - |
| DDL | 3.97 | 13.75 | 0.71 | 0.58 | 0.64 | - | - |
| SYSU-ISE | 5.45 | 10.65 | 0.6 | 0.48 | 0.52 | - | - |
| Active Exploration (Pre-explore) | 3.3 | 9.85 | 0.77 | 0.68 | 0.7 | - | - |
| zq | 4.32 | 13.43 | 0.69 | 0.55 | 0.61 | - | - |
| trysth | 4.49 | 12.22 | 0.67 | 0.54 | 0.59 | - | - |
| DCMT | 4.96 | 12.51 | 0.61 | 0.48 | 0.54 | - | - |
| lxyict | 3.61 | 14.6 | 0.77 | 0.59 | 0.7 | - | - |
| test-sf | 4.57 | 10.99 | 0.65 | 0.5 | 0.57 | - | - |
| single-run | 3.87 | 13.11 | 0.72 | 0.59 | 0.65 | Vision-Language Navigation with Random Environmental Mixup | |
| HOC | 3.38 | 15.47 | 0.79 | 0.59 | 0.71 | - | - |
| Self-Monitoring Navigation Agent (no beam search; Progress Inference) | 5.67 | 18.04 | 0.59 | 0.35 | 0.48 | Self-Monitoring Navigation Agent via Auxiliary Progress Estimation | |
| Khanh Nguyen | 6.46 | 13.08 | 0.45 | 0.3 | 0.37 | - | - |
| single-run | 5.37 | 10.0 | 0.59 | 0.5 | 0.53 | - | - |
| ADad | 3.31 | 786.35 | 0.99 | 0.01 | 0.69 | - | - |
| reward-vln | 4.65 | 12.19 | 0.62 | 0.52 | 0.56 | - | - |
| SEA features + AuxRN (single-run) | 4.71 | 10.31 | 0.64 | 0.55 | 0.59 | - | - |
| Single-Run, No Pre-Explore | 3.73 | 15.89 | 0.73 | 0.6 | 0.66 | - | - |
| Reinforced Cross-Modal Matching (single trajectory; NO beam search) | 6.12 | 11.97 | 0.5 | 0.38 | 0.43 | - | - |
| Lily | 3.44 | 16.14 | 0.79 | 0.6 | 0.72 | - | - |
0 of 145 row(s) selected.