Research map / Detection and segmentation / Open-set and few-shot detection
LMSeg: Language-guided Multi-dataset Segmentation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 LMSeg instead of multi-head training | Multi-dataset semantic segmentation · Four semantic datasets (average) | mIoU: 53.76 -> 54.04 | Helped |
| Use LMSeg instead of manual-taxonomy training | Multi-dataset panoptic segmentation · Three panoptic datasets (average) | PQ: 38.53 -> 42.40 | Helped |
| Initialize image and text encoders with CLIP rather than using ImageNet image initialization (ablation) | LMSeg · Four semantic datasets (average) | mIoU: 50.30 to 52.75 | Helped |
| Place visual cross-attention before text cross-attention (ablation) | LMSeg category-guided decoding · Four semantic datasets (average) | mIoU: 52.95 -> 53.41 | Helped |
| Add category-guided decoding (ablation) | LMSeg · Four semantic datasets (average) | mIoU: 52.91 to 53.41 | Helped |
| Add category-guided decoding (ablation) | LMSeg · Three panoptic datasets (average) | PQ: 34.38 to 37.88 | Helped |
| Add dataset-aware augmentation to category-guided decoding (ablation) | LMSeg · Four semantic datasets (average) | mIoU: 53.41 to 54.04 | Helped |
| Add dataset-aware augmentation to category-guided decoding (ablation) | LMSeg · Three panoptic datasets (average) | PQ: 37.88 to 38.91 | Helped |
Closest papers by meaning
- Automated Label Unification for Multi-Dataset Semantic Segmentation with GNNs (NeurIPS 2024 · similarity 0.88)
- DynAlign: Unsupervised Dynamic Taxonomy Alignment for Cross-Domain Segmentation (ICLR 2025 · similarity 0.86)
- Excluding the Impossible for Open Vocabulary Semantic Segmentation (AAAI 2025 · similarity 0.85)
- USE: Universal Segment Embeddings for Open-Vocabulary Image Segmentation (CVPR 2024 · similarity 0.85)
- Label-Efficient Few-Shot Semantic Segmentation with Unsupervised Meta-Training (AAAI 2024 · similarity 0.85)
- Instruction-guided Multi-Granularity Segmentation and Captioning with Large Multimodal Model (AAAI 2025 · similarity 0.85)
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.