Research map / Detection and segmentation / Open-set and few-shot detection
Urban Socio-Semantic Segmentation with Vision-Language Reasoning
Yu Wang, Yi Wang, Rui Dai, Yujie Wang, Kaikui Liu, Xiangxiang Chu, Yansheng Li
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 reflection (two-stage render-and-refine instead of single-stage) (ablation) | SocioReasoner · SocioSeg test (All dataset) | gIoU: 48.5 -> 52.8 | Improved |
| add refinement stage after Stage-1 localization (ablation) | SocioReasoner · SocioSeg test (All dataset) | gIoU: 50.8 -> 52.8 | Improved |
| add refinement stage after Stage-1 localization (ablation) | SocioReasoner · SocioSeg test (All dataset) | F1: 57.5 -> 59.7 | Improved |
| use 2 points instead of 1 in refinement (ablation) | SocioReasoner · SocioSeg test (All dataset) | gIoU: 51.2 -> 52.8 | Improved |
| use 3 points instead of 2 in refinement (ablation) | SocioReasoner · SocioSeg test (All dataset) | gIoU: 52.8 -> 52.3 gIoU; cIoU 47.9 -> 48.9 | Mixed |
| train with RL instead of SFT (ablation) | SocioReasoner · SocioSeg OOD (Map Style) | gIoU: 42.0 -> 49.1 | Improved |
| train with RL instead of SFT (ablation) | SocioReasoner · SocioSeg OOD (New Region) | gIoU: 32.3 -> 43.4 | Improved |
| train with RL instead of SFT (ablation) | SocioReasoner · SocioSeg test (In-domain) | gIoU: 51.4 -> 52.8 | Improved |
| train with RL instead of SFT (ablation) | VisionReasoner · SocioSeg OOD (New Region) | gIoU: 24.7 -> 34.4 | Improved |
| train with RL instead of SFT (ablation) | VisionReasoner · SocioSeg OOD (Map Style) | gIoU: 40.9 -> 44.4 | Improved |
| replace VisionReasoner single-stage with SocioReasoner two-stage | VisionReasoner · SocioSeg test (All dataset) | cIoU: 44.0 -> 47.9 | Improved |
| remove frozen CLIP-based open-vocabulary approach (SegEarth-OV) vs trained reasoning VLM | SegEarth-OV · SocioSeg test (All dataset) | gIoU: 3.7 vs 52.8 | Hurt |
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.
- SegEarth-OV — SegEarth-OV: Towards Training-Free Open-Vocabulary Segmentation for Remote Sensing Images (CVPR 2025) · better on 2 of 2 shared results
- Segformer — SegFormer: A Topic Segmentation Model with Controllable Range of Attention (AAAI 2023) · better on 2 of 2 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 Curriculum Domain Adaptation Approach to the Semantic Segmentation of Urban Scenes (TPAMI 2019 · similarity 0.88)
- CityLens: Evaluating Large Vision-Language Models for Urban Socioeconomic Sensing (ICLR 2026 · similarity 0.87)
- The Cityscapes Dataset for Semantic Urban Scene Understanding (CVPR 2016 · similarity 0.87)
- LISAt: Language-Instructed Segmentation Assistant for Satellite Imagery (NeurIPS 2025 · similarity 0.87)
- Exploring Simple Open-Vocabulary Semantic Segmentation (CVPR 2025 · similarity 0.86)
- UV-SAM: Adapting Segment Anything Model for Urban Village Identification (AAAI 2024 · 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.
