Research map / Detection and segmentation / Open-set and few-shot detection
Beyond Text: Visual Description Assembly by Probabilistic Model for CLIP-based Weakly Supervised Semantic Segmentation
Xianglin Qiu, Jian Wang, Xiaolei Wang, Zhen Zhang, Jimin Xiao
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 VDAF (visual description assembly and fusion) with the zero-shot filter (ablation) | CLIP text-template baseline · PASCAL VOC 2012 val | mIoU (%): 75.8 -> 78.4 | Improved |
| add DSE (decoder semantic enhancement) alone (ablation) | CLIP text-template baseline · PASCAL VOC 2012 val | mIoU (%): 75.8 -> 76.5 | Improved |
| add the zero-shot reliable-prototype filter (ablation) | VDA (VDAF + DSE) · PASCAL VOC 2012 val | mIoU (%): 75.3 -> 78.7 | Improved |
| combine VDAF and DSE (ablation) | VDA · PASCAL VOC 2012 val | mIoU (%): 78.4 -> 78.7 | Improved |
| use 8 attributes per class instead of 3 (ablation) | VDA intra-class GMM · PASCAL VOC 2012 val | mIoU (%): 77.9 -> 78.7 | Improved |
| increase attributes per class from 8 to 15 (ablation) | VDA intra-class GMM · PASCAL VOC 2012 val | mIoU (%): 78.7 -> 78.4 | Hurt |
| use density-based dynamic fusion instead of a fixed weight (ablation) | VDA query prototype · PASCAL VOC 2012 val | mIoU (%): 76.5 (alpha=0) / 77.6 (alpha=1) -> 78.7 | Improved |
| use visual prototype only instead of text prototype only (ablation) | VDA query prototype · PASCAL VOC 2012 val | mIoU (%): 76.5 -> 77.6 | Improved |
| add CRF post-processing (ablation) | VDA · PASCAL VOC 2012 val | mIoU (%): 78.7 -> 79.9 | Improved |
| replace static text query with assembled visual query in single-stage WSSS | ExCEL · MS COCO 2014 val | mIoU (%): 50.3 -> 51.4 | Improved |
Closest papers by meaning
- Exploring CLIP's Dense Knowledge for Weakly Supervised Semantic Segmentation (CVPR 2025 · similarity 0.92)
- Toward Modality Gap: Vision Prototype Learning for Weakly-supervised Semantic Segmentation with CLIP (AAAI 2025 · similarity 0.91)
- Object-centric binding in Contrastive Language-Image Pretraining (NeurIPS 2025 · similarity 0.90)
- Rethinking Prior Information Generation with CLIP for Few-Shot Segmentation (CVPR 2024 · similarity 0.90)
- Building Vision-Language Models on Solid Foundations with Masked Distillation (CVPR 2024 · similarity 0.89)
- SmartCLIP: Modular Vision-language Alignment with Identification Guarantees (CVPR 2025 · 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.
