Research map / Detection and segmentation / Open-set and few-shot detection
SCORE: Scene Context Matters in Open-Vocabulary Remote Sensing Instance Segmentation
Shiqi Huang, Shuting He, Huaiyuan Qin, Bihan Wen
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 Region-Aware Integration (RAI) (ablation) | SCORE · OV RS instance seg (train iSAID, average) | mAP: 33.25 -> 36.91 | Improved |
| add Global Context Adaptation (GCA) (ablation) | SCORE · OV RS instance seg (train iSAID, average) | mAP: 33.25 -> 36.67 | Improved |
| add both RAI and GCA (ablation) | SCORE · OV RS instance seg (train iSAID, average) | mAP: 33.25 -> 38.68 | Improved |
| add both RAI and GCA (ablation) | SCORE · OV RS instance seg (train SIOR, average) | mAP: 28.36 -> 32.59 | Improved |
| use RemoteCLIP instead of general CLIP for scene context (ablation) | SCORE · OV RS instance seg (train iSAID) | mAP: 64.03 -> 67.59 | Improved |
| use regional context instead of [CLS] token in RAI (ablation) | SCORE · OV RS instance seg (train iSAID) | mAP: 67.06 -> 67.59 | Improved |
| use regional context instead of patch embeddings in RAI (ablation) | SCORE · OV RS instance seg (train iSAID) | mAP: 66.14 -> 67.59 | Improved |
| inject global context by MHA instead of addition (ablation) | SCORE GCA · OV RS instance seg (train iSAID) | mAP: 65.72 -> 67.59 | Improved |
| inject global context by MHA instead of concatenation (ablation) | SCORE GCA · OV RS instance seg (train iSAID) | mAP: 47.53 -> 67.59 | Improved |
| use general CLIP instead of RemoteCLIP for out-vocabulary classification (ablation) | SCORE · OV RS instance seg (train iSAID) | mAP: 66.05 -> 67.59 | Improved |
| use SCORE instead of FC-CLIP baseline | FC-CLIP · OV RS instance seg (train iSAID, average) | mAP: 33.24 -> 38.68 | Improved |
| apply SCORE to semantic segmentation instead of GSNet | GSNet · OV RS semantic seg (LandDiscover50K) | mIoU: 28.63 -> 29.76 | Mixed |
Closest papers by meaning
- Towards Open-Vocabulary Remote Sensing Image Semantic Segmentation (AAAI 2025 · similarity 0.91)
- ZoRI: Towards Discriminative Zero-Shot Remote Sensing Instance Segmentation (AAAI 2025 · similarity 0.90)
- Locate Anything on Earth: Advancing Open-Vocabulary Object Detection for Remote Sensing Community (AAAI 2025 · similarity 0.89)
- InstructSAM: A Training-free Framework for Instruction-Oriented Remote Sensing Object Recognition (NeurIPS 2025 · similarity 0.88)
- SkySense-O: Towards Open-World Remote Sensing Interpretation with Vision-Centric Visual-Language Modeling (CVPR 2025 · similarity 0.88)
- SegEarth-OV: Towards Training-Free Open-Vocabulary Segmentation for Remote Sensing Images (CVPR 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.
