Research map / Vision-language and multimodal / Robot action and navigation
Learning Disentanglement with Decoupled Labels for Vision-Language Navigation
Wenhao Cheng, Xingping Dong, Salman Khan, Jianbing Shen
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add DDL framework | HAMT* · R2R validation unseen | SR: 65.6 -> 67.9 | Improved |
| add DDL framework | HAMT* · R2R test unseen | SR: 64.4 -> 66.3 | Improved |
| add DDL framework | OAAM* · R2R validation unseen | SR: 54.4 -> 57.6 | Improved |
| add DDL framework | VLN-BERT* · R2R validation unseen | SR: 62.2 -> 64.8 | Improved |
| add DDL framework | VLN-BERT* · R4R validation unseen | CLS: 39.4 -> 43.6 | Improved |
| add annotated LAR2R labels to regularize language attention in IL (ablation) | OAAM · R2R validation unseen | SR: 50.2 -> 50.8 | Improved |
| add landmark and action pseudo-labels in RL phase (ablation) | OAAM · R2R validation unseen | SR: 50.8 -> 51.5 | Improved |
| add DLS pseudo-labels on back-translation augmented data (ablation) | OAAM · R2R validation unseen | SR: 54.4 -> 57.6 | Improved |
| use random language labels instead of no labels (ablation) | OAAM · R2R validation unseen | SR: 50.2 -> 47.9 | Hurt |
| use fine-grained LAR2R labels instead of FGR2R sub-instruction labels (ablation) | OAAM · R2R validation unseen | SR: 49.6 -> 50.8 | Improved |
| use all-ones average labels instead of LAR2R labels (ablation) | OAAM · R2R validation unseen | SR: 50.8 -> 48.7 | Hurt |
Closest papers by meaning
- Towards Learning a Generic Agent for Vision-and-Language Navigation via Pre-Training (CVPR 2020 · similarity 0.89)
- Learning from Unlabeled 3D Environments for Vision-and-Language Navigation (ECCV 2022 · similarity 0.88)
- NavCoT: Boosting LLM-Based Vision-and-Language Navigation via Learning Disentangled Reasoning (TPAMI 2025 · similarity 0.88)
- Transferable Representation Learning in Vision-and-Language Navigation (ICCV 2019 · similarity 0.88)
- Counterfactual Cycle-Consistent Learning for Instruction Following and Generation in Vision-Language Navigation (CVPR 2022 · similarity 0.88)
- DecoVLN: Decoupling Observation, Reasoning, and Correction for Vision-and-Language Navigation (CVPR 2026 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
