Research map / Point clouds / Point cloud classification
CityRefer: Geography-aware 3D Visual Grounding Dataset on City-scale Point Cloud Data
Taiki Miyanishi, Fumiya Kitamori, Shuhei Kurita, Jungdae Lee, Motoaki Kawanabe, Nakamasa Inoue
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 landmark features and landmark names to baseline | Baseline · CityRefer overall | Acc@0.25IoU: 0.312 -> 0.320 | Improved |
| add landmark information | Baseline · CityRefer Car | Acc@0.25IoU: 0.282 -> 0.298 | Improved |
| add landmark information | Baseline · CityRefer Building | Acc@0.25IoU: 0.255 -> 0.255 | No clear effect |
| add landmark information | Baseline · CityRefer Parking | Acc@0.25IoU: 0.835 -> 0.853 | Improved |
| learned baseline versus random guess | Baseline + Land · CityRefer overall | Acc@0.25IoU: 0.101 -> 0.320 | Improved |
| remove RGB color from points (ablation) | Baseline + Land · CityRefer overall | Acc@0.25IoU: 0.320 -> 0.303 | Hurt |
| remove RGB color from points (ablation) | Baseline + Land · CityRefer Building | Acc@0.25IoU: 0.255 -> 0.235 | Hurt |
| remove landmark names (ablation) | Baseline + Land · CityRefer overall | Acc@0.25IoU: 0.320 -> 0.310 | Hurt |
| use 2048 instance points instead of 1024 (ablation) | Baseline + Land · CityRefer overall | Acc@0.25IoU: 0.320 -> 0.302 | Hurt |
| use 512 instance points instead of 1024 (ablation) | Baseline + Land · CityRefer overall | Acc@0.25IoU: 0.320 -> 0.314 | Hurt |
| use 512 instance points instead of 1024 (ablation) | Baseline + Land · CityRefer Parking | Acc@0.25IoU: 0.853 -> 0.879 | Improved |
Closest papers by meaning
- SensatUrban: Learning Semantics from Urban-Scale Photogrammetric Point Clouds (IJCV 2022 · similarity 0.90)
- CityAnchor: City-scale 3D Visual Grounding with Multi-modality LLMs (ICLR 2025 · similarity 0.90)
- WildRefer: 3D Object Localization in Large-scale Dynamic Scenes with Multi-modal Visual Data and Natural Language (ECCV 2024 · similarity 0.89)
- Towards Semantic Segmentation of Urban-Scale 3D Point Clouds: A Dataset, Benchmarks and Challenges (CVPR 2021 · similarity 0.89)
- CitySTAR: Agent-Driven Structured and Topology-Aware Reasoning for Open-Vocabulary Urban 3D Grounding (NeurIPS 2026 · similarity 0.87)
- USCILab3D: A Large-scale, Long-term, Semantically Annotated Outdoor Dataset (NeurIPS 2024 · 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.
