Research map / Vision-language and multimodal / Robot action and navigation
Learning Vision-and-Language Navigation from YouTube Videos
Kunyang Lin, Peihao Chen, Diwei Huang, Thomas H. Li, Mingkui Tan, Chuang Gan
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 |
|---|---|---|---|
| use entropy-based frame selection instead of random sampling (ablation) | YouTube-VLN pre-training · R2R val unseen | SR (%): 64.41 -> 66.70 | Improved |
| use entropy-based frame selection instead of temporal difference (ablation) | YouTube-VLN pre-training · R2R val unseen | SR (%): 65.30 -> 66.70 | Improved |
| use YouTube video frames instead of Airbnb images as data source (ablation) | VLN pre-training · R2R val unseen | SR (%): +0.89% SR | Improved |
| use reasonable chronological trajectories instead of shuffled trajectories (ablation) | VLN pre-training · R2R val unseen | SR (%): +1.27% SR | Improved |
| fill instructions with pseudo-labeled actions instead of random action words (ablation) | action-aware instruction generator · R2R val unseen | SR (%): +1.10% SR (val unseen), +0.30% (val seen) | Improved |
| add trajectory judgment pretext task (ablation) | Lily pre-training · R2R val seen / val unseen | SR (%): +2.05% val seen, +0.60% val unseen | Improved |
| pre-train DUET with YouTube-VLN and trajectory judgment | DUET · REVERIE val unseen | SR (%): 46.98 -> 48.11 | Improved |
| pre-train DUET with YouTube-VLN and trajectory judgment | DUET · R2R val unseen (generative) | SR (%): 72 -> 74 | Improved |
| pre-train on YouTube-VLN then fine-tune on one environment | Lily · R2R val unseen | SR (%): 22.43 -> 50.86 (vs VLN-BERT) | Improved |
| Lily versus Airbert on R2R val unseen | Lily · R2R val unseen (discriminative) | SR (%): 68.67 -> 70.00 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Airbert — Airbert: In-Domain Pretraining for Vision-and-Language Navigation (ICCV 2021) · better on 6 of 6 shared results
- DUET — DUET: Cross-Modal Semantic Grounding for Contrastive Zero-Shot Learning (AAAI 2023) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- A New Path: Scaling Vision-and-Language Navigation With Synthetic Instructions and Imitation Learning (CVPR 2023 · similarity 0.91)
- Towards Learning a Generic Agent for Vision-and-Language Navigation via Pre-Training (CVPR 2020 · similarity 0.90)
- RoomTour3D: Geometry-Aware Video-Instruction Tuning for Embodied Navigation (CVPR 2025 · similarity 0.90)
- Learning from Unlabeled 3D Environments for Vision-and-Language Navigation (ECCV 2022 · similarity 0.90)
- Curriculum Learning for Vision-and-Language Navigation (NeurIPS 2021 · similarity 0.90)
- Human-Aware Vision-and-Language Navigation: Bridging Simulation to Reality with Dynamic Human Interactions (NeurIPS 2024 · 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.
