Research map / Vision-language and multimodal / Robot action and navigation
PanoGen: Text-Conditioned Panoramic Environment Generation for Vision-and-Language Navigation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 speaker-data pre-training with mPLUG speaker on PanoGen environments (ablation) | DUET-CLIP · CVDN val unseen | GP: 5.50 -> 5.93 | Helped |
| add speaker-data pre-training with mPLUG speaker (ablation) | DUET-CLIP · R2R val unseen | SR: 72.84 -> 74.20 | Helped |
| add speaker-data pre-training with mPLUG speaker (ablation) | DUET-CLIP · R2R val unseen | SPL: 63.37 -> 62.81 | Hurt |
| add speaker-data pre-training with mPLUG speaker (ablation) | DUET-CLIP · R4R val unseen | SPL: 41.94 -> 42.52 | Helped |
| use mPLUG speaker instead of EnvDrop speaker (ablation) | PanoGen pre-training · CVDN val unseen | GP: 5.81 -> 5.93 | Helped |
| use mPLUG speaker instead of EnvDrop speaker (ablation) | PanoGen pre-training · R4R val unseen | SPL: 40.33 -> 42.52 | Helped |
| add observation replacement with PanoGen panoramas in fine-tuning (ablation) | DUET-CLIP · R2R val unseen | SR: 72.84 -> 74.41 | Helped |
| add observation replacement in fine-tuning (ablation) | DUET-CLIP · CVDN val unseen | GP: 5.50 -> 5.63 | Helped |
| add observation replacement in fine-tuning (ablation) | DUET-CLIP · R4R val unseen | SPL: 41.94 -> 44.25 | Helped |
| increase replacement ratio from 0.3 to 0.7 (ablation) | PanoGen+Replace · R2R val unseen | SR: 74.41 -> 72.33 | Hurt |
| replacement ratio 0.1 (ablation) | DUET-CLIP · R2R val unseen | SR: 72.84 -> 72.84 | No clear effect |
| increase number of PanoGen scans used in fine-tuning from 10 to 122 (ablation) | DUET-CLIP · R2R val unseen | SPL: 62.48 -> 64.27 | Helped |
Closest papers by meaning
- PanoWorld: Towards Spatial Supersensing in 360◦ Panorama World (NeurIPS 2026 · similarity 0.88)
- PanoEnv: Exploring 3D Spatial Intelligence in Panoramic Environments with Reinforcement Learning (CVPR 2026 · similarity 0.88)
- DiffPano: Scalable and Consistent Text to Panorama Generation with Spherical Epipolar-Aware Diffusion (NeurIPS 2024 · similarity 0.87)
- PanoWan: Lifting Diffusion Video Generation Models to 360$^\circ$ with Latitude/Longitude-aware Mechanisms (NeurIPS 2025 · similarity 0.87)
- Conditional Panoramic Image Generation via Masked Autoregressive Modeling (NeurIPS 2025 · similarity 0.87)
- General Scene Adaptation for Vision-and-Language Navigation (ICLR 2025 · similarity 0.86)
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.