Research map / Detection and segmentation / Open-set and few-shot detection
Open-Vocabulary RGB-Thermal Semantic Segmentation
GuoQiang Zhao, JunJie Huang, Xiaoyun Yan, Zhaojing Wang, Junwei Tang, Yangjun Ou, Xinrong Hu, Tao Peng
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 Thermal Information Prompt (TIP) module at every Transformer layer instead of additive fusion (ablation) | OpenRSS (SAM-H) · Freiburg Thermal | mIoU: 3.90 -> 27.55 | Improved |
| add Thermal Information Prompt (TIP) module at every Transformer layer instead of additive fusion (ablation) | OpenRSS (SAM-H) · MFNet | mIoU: 10.20 -> 22.81 | Improved |
| insert TIP at every layer rather than only before the first layer (ablation) | OpenRSS (SAM-H) · MFNet | mIoU: 19.98 -> 22.81 | Improved |
| insert TIP at every layer rather than only before the first layer (ablation) | OpenRSS (SAM-H) · Freiburg Thermal | mIoU: 25.44 -> 27.55 | Improved |
| use correlative self-attention instead of multi-head self-attention inside TIP (ablation) | OpenRSS TIP module · MCubeS | mIoU: 21.72 -> 28.58 | Improved |
| use correlative self-attention instead of spatial fovea attention inside TIP (ablation) | OpenRSS TIP module · Freiburg Thermal | mIoU: 18.58 -> 27.55 | Improved |
| use correlative self-attention instead of multi-head self-attention inside TIP (ablation) | OpenRSS TIP module · MFNet | mIoU: 22.54 -> 22.81 | Improved |
| use dynamic LoRA with four bypasses instead of fixed rank 1 (ablation) | OpenRSS (SAM-H) · Freiburg Thermal | mIoU: 19.57 -> 27.55 | Improved |
| use dynamic LoRA instead of best fixed rank (8) (ablation) | OpenRSS (SAM-H) · MFNet | mIoU: 22.51 -> 22.81 | Improved |
| use dynamic LoRA instead of fixed rank 4 (ablation) | OpenRSS (SAM-H) · MCubeS | mIoU: 27.73 -> 28.58 | Improved |
| scale SAM backbone from B to H | OpenRSS · Freiburg Thermal | mIoU: 8.14 -> 27.55 | Improved |
Closest papers by meaning
- Towards Open-Vocabulary Remote Sensing Image Semantic Segmentation (AAAI 2025 · similarity 0.87)
- Leveraging Depth and Language for Open-Vocabulary Domain-Generalized Semantic Segmentation (NeurIPS 2025 · similarity 0.87)
- SED: A Simple Encoder-Decoder for Open-Vocabulary Semantic Segmentation (CVPR 2024 · similarity 0.87)
- Open-Vocabulary Semantic Segmentation with Image Embedding Balancing (CVPR 2024 · similarity 0.87)
- Effective SAM Combination for Open-Vocabulary Semantic Segmentation (CVPR 2025 · similarity 0.87)
- Emergent Open-Vocabulary Semantic Segmentation from Off-the-shelf Vision-Language Models (CVPR 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.
