Research map / Graph learning / Knowledge and temporal graphs
STAR: A First-Ever Dataset and a Large-Scale Benchmark for Scene Graph Generation in Large-Size Satellite Imagery
Yansheng Li, Linlin Wang, Tingzhu Wang, Xue Yang, Junwei Luo, Qi Wang, Youming Deng, Wenbin Wang, Xian Sun, Haifeng Li, Bo Dang, Yongjun Zhang and 2 more
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 |
|---|---|---|---|
| use HOD-Net holistic detection instead of cropping | Faster R-CNN Swin-L · STAR OBD HBB | mAP: 42.6 -> 53.2 | Improved |
| use HOD-Net instead of resizing | Faster R-CNN Swin-L · STAR OBD HBB | mAP: 33.7 -> 53.2 | Improved |
| use HOD-Net instead of cropping | Oriented R-CNN Swin-L · STAR OBD OBB | mAP: 41.9 -> 55.9 | Improved |
| use HOD-Net instead of resizing | Oriented R-CNN Swin-L · STAR OBD OBB | mAP: 30.9 -> 55.9 | Improved |
| swap ResNet50 for Swin-L backbone | HOD-Net · STAR OBD OBB | mAP: 43.6 -> 55.9 | Improved |
| use PPG pair pruning instead of random pairs (ablation) | RPCM · STAR SGG PredCls OBB | HMR@2000: 40.15 -> 51.51 | Improved |
| use PPG instead of ABBS statistics pruning (ablation) | RPCM · STAR SGG PredCls OBB | HMR@2000: 41.51 -> 51.51 | Improved |
| use PPG instead of random pruning (ablation) | RPCM · STAR SGG SGDet OBB | HMR@2000: 9.60 -> 16.96 | Improved |
| add object context augmentation (OCA) (ablation) | RPCM · STAR SGG PredCls | HMR@2000: 46.27/48.08 -> 47.87/48.83 | Improved |
| add relationship context augmentation (RCA) (ablation) | RPCM · STAR SGG PredCls | HMR@2000: 47.87/48.83 -> 50.23/51.51 | Improved |
| use prototype matching instead of linear classifier (ablation) | RPCM · STAR SGG PredCls | mMR@2000: 35.80 -> 42.30 | Improved |
| use prototype matching instead of linear classifier (ablation) | RPCM · STAR SGG PredCls | MR@2000: 68.18 -> 65.86 | Hurt |
Closest papers by meaning
- Adaptive Visual Scene Understanding: Incremental Scene Graph Generation (NeurIPS 2024 · similarity 0.87)
- Scene Graph Generation Strategy with Co-occurrence Knowledge and Learnable Term Frequency (ICML 2024 · similarity 0.87)
- HiKER-SGG: Hierarchical Knowledge Enhanced Robust Scene Graph Generation (CVPR 2024 · similarity 0.87)
- GG-SSMs: Graph-Generating State Space Models (CVPR 2025 · similarity 0.86)
- DSGG: Dense Relation Transformer for an End-to-end Scene Graph Generation (CVPR 2024 · similarity 0.86)
- Adaptive Self-training Framework for Fine-grained Scene Graph Generation (ICLR 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.
