Research map / Vision-language and multimodal / CLIP and zero-shot
Segmenting Visuals With Querying Words: Language Anchors For Semi-Supervised Image Segmentation
Numair Nadeem, Saeed Anwar, Muhammad Asad, Abdul Bais
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 |
|---|---|---|---|
| replace TQDM baseline with full HVLFormer | TQDM (SigLIP2) · Pascal VOC (92 labels) | mIoU: 76.1 -> 89.4 | Improved |
| add hierarchical query generation (ablation) | HVLFormer SigLIP2 · COCO (232 labels) | mIoU: 41.9 -> 45.3 | Improved |
| add semantic relevance estimation (SRE) (ablation) | HVLFormer SigLIP2 · Pascal VOC (92 labels) | mIoU: 79.4 -> 80.5 | Improved |
| add pixel-text refinement module (PTRM) (ablation) | HVLFormer SigLIP2 · Pascal VOC (92 labels) | mIoU: 80.5 -> 84.1 | Improved |
| add full CMCR (mask+class+align) (ablation) | HVLFormer SigLIP2 · COCO (232 labels) | mIoU: 54.8 -> 59.4 | Improved |
| use soft weighting by s_k instead of no SRE (ablation) | HVLFormer SigLIP2 · COCO (232 labels) | mIoU: 57.8 -> 59.4 | Improved |
| use fixed threshold s_k>0.5 instead of soft weighting (ablation) | HVLFormer SigLIP2 · Pascal VOC (92 labels) | mIoU: 89.4 -> 88.7 | Hurt |
| add dataset attributes to learnable prompt (ablation) | HVLFormer SigLIP2 · Cityscapes (100 labels) | mIoU: 78.3 -> 79.6 | Improved |
| use learnable instead of fixed prompt (class name only) (ablation) | HVLFormer SigLIP2 · COCO (232 labels) | mIoU: 54.2 -> 58.5 | Improved |
| increase hierarchy levels E from 1 to 3 (ablation) | HVLFormer SigLIP2 · Pascal VOC (92 labels) | mIoU: 87.3 -> 88.9 | Improved |
| replace cross-attention with PTRM (ablation) | HVLFormer SigLIP2 · COCO (232 labels) | mIoU: 57.2 -> 59.4 | Improved |
| replace vision-only UniMatch consistency with CMCR (ablation) | HVLFormer SigLIP2 · Pascal VOC (92 labels) | mIoU: 85.2 -> 89.4 | Improved |
Closest papers by meaning
- SemiDAViL: Semi-supervised Domain Adaptation with Vision-Language Guidance for Semantic Segmentation (CVPR 2025 · similarity 0.90)
- DH-Set: Improving Vision-Language Alignment with Diverse and Hybrid Set-Embeddings Learning (CVPR 2025 · similarity 0.90)
- Locality Alignment Improves Vision-Language Models (ICLR 2025 · similarity 0.90)
- Emergent Open-Vocabulary Semantic Segmentation from Off-the-shelf Vision-Language Models (CVPR 2024 · similarity 0.90)
- DINOv2 Meets Text: A Unified Framework for Image- and Pixel-Level Vision-Language Alignment (CVPR 2025 · similarity 0.89)
- Revisiting the Role of Language Priors in Vision-Language Models (ICML 2024 · similarity 0.89)
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.
