Research map / 3D vision / Human pose estimation
Embodied Question Answering in Photorealistic Environments With Point Cloud Perception
Check 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 |
|---|---|---|---|
| Use inflection-weighted rather than unweighted behavior-cloning loss (ablation) | Memory navigation models · MP3D-EQA test, T_-30 | d_T: 1.330 -> 0.830 | Helped |
| Use inflection weighting (ablation) | M+PC · MP3D-EQA test, T_-30 | IoU_T: 0.114 -> 0.114 | Helped |
| Add memory to a reactive navigator (ablation) | Navigation models · MP3D-EQA test, T_-30 | d_T: 1.330 -> 0.830 | Helped |
| Add RGB perception (ablation) | Memory navigator · MP3D-EQA test, T_-30 | d_T: 0.830 -> 1.157 | Hurt |
| Add RGB perception (ablation) | Memory navigator · MP3D-EQA test, T_-30 | IoU_T: 0.091 -> 0.136 | Helped |
| Add point-cloud perception (ablation) | Memory navigator · MP3D-EQA test, T_-30 | %_collision: 10.989 -> 9.169 | Helped |
| Add both point-cloud and RGB perception (ablation) | Memory navigator · MP3D-EQA test, T_-30 | IoU_T: 0.091 -> 0.179 | Helped |
| Add the question input (ablation) | M+PC · MP3D-EQA test, T_-30 | d_T: 1.020 -> 1.030 | Hurt |
| Add the question input (ablation) | M+PC+RGB · MP3D-EQA test, T_-30 | d_T: 0.940 -> 1.044 | Hurt |
| Jointly train RGB encoder on segmentation, depth, and autoencoding (ablation) | ResNet50 · MP3D-EQA validation | MIOU: 0.002 -> 0.078 | Helped |
| Use spatial attention rather than standard attention for VQA (ablation) | RGB question-answering model · MP3D-EQA validation | Top-1 (%): 40.0 -> 46.2 | Helped |
| Combine RGB and point-cloud inputs (ablation) | Memory navigator · MP3D-EQA test, T_-30 | %_collision: 9.169 -> 9.574 | Mixed |
Closest papers by meaning
- D3D-VLP: Dynamic 3D Vision-Language-Planning Model for Embodied Grounding and Navigation (CVPR 2026 · similarity 0.88)
- NavBench: Probing Multimodal Large Language Models for Embodied Navigation (NeurIPS 2025 · similarity 0.87)
- Image2Sim: Scaling Embodied Navigation via Generative Neural Simulator (NeurIPS 2026 · similarity 0.87)
- Multi-modal Situated Reasoning in 3D Scenes (NeurIPS 2024 · similarity 0.87)
- Embodied Scene Understanding for Vision Language Models via MetaVQA (CVPR 2025 · similarity 0.87)
- Spatial Understanding from Videos: Structured Prompts Meet Simulation Data (NeurIPS 2025 · similarity 0.87)
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.