Research map / Large language models / Language, safety and interpretability
FlowSeg: Dynamic Semantic Guidance for LLM-Conditioned Segmentation
Zekang Zhang, Guangyu Gao, Youyun Tang, WU CHENGJING, Xiaochao Qu, Chi Harold Liu, Jianbo Jiao, Yunchao Wei, Luoqi Liu, Ting Liu
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 Semantic Cross-Attention with Adaptive Fusion (SR) (ablation) | FlowSeg baseline · RefCOCO/+/g val | avg cIoU: 82.4 -> 82.9 | Improved |
| add Condition Refinement on top of SR (ablation) | FlowSeg · RefCOCO/+/g val | avg cIoU: 82.9 -> 83.9 | Improved |
| add Boundary-Aware Refinement on top of SR+CR (ablation) | FlowSeg · RefCOCO/+/g val | avg cIoU: 83.9 -> 84.2 | Improved |
| add boundary-aware mask refinement (ablation) | FlowSeg · RefCOCO val | gBIoU: 70.6 -> 72.2 | Improved |
| upgrade X-SAM LLM backbone to Qwen3-4B (ablation) | X-SAM · RefCOCOg val | cIoU: 82.5 -> 84.1 | Improved |
| replace X-SAM decoder with FlowSeg using same Phi-3 backbone (ablation) | FlowSeg (Phi-3) · RefCOCOg val | cIoU: 82.5 -> 85.9 | Improved |
| use bidirectional semantic flow on X-SAM failure cases (cIoU<0.5) | FlowSeg · RefCOCO/+/g val failure subsets | avg cIoU: 4.6 -> 49.2 | Improved |
| compare candidate quality of FlowSeg vs X-SAM (oracle upper bound) (ablation) | FlowSeg · RefCOCO val | oracle cIoU: 91.31 -> 91.47 | No clear effect |
| FlowSeg vs HyperSeg on ReasonSeg val | FlowSeg · ReasonSeg val | cIoU: 56.7 -> 49.2 | Hurt |
Closest papers by meaning
- Mask-Adapter: The Devil is in the Masks for Open-Vocabulary Segmentation (CVPR 2025 · similarity 0.88)
- OneRef: Unified One-tower Expression Grounding and Segmentation with Mask Referring Modeling (NeurIPS 2024 · similarity 0.87)
- Semantic Routing: Exploring Multi-Layer LLM Feature Weighting for Diffusion Transformers (NeurIPS 2026 · similarity 0.87)
- SAMTok: Representing Any Mask with Two Words (CVPR 2026 · similarity 0.87)
- MaskViM: Domain Generalized Semantic Segmentation with State Space Models (AAAI 2025 · similarity 0.86)
- Mask Grounding for Referring Image Segmentation (CVPR 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.
