Research map / Vision-language and multimodal / CLIP and zero-shot
Probabilistic Prototype Calibration of Vision-language Models for Generalized Few-shot Semantic Segmentation
Jie Liu, Jiayi Shen, Pan Zhou, Jan-Jakob Sonke, Efstratios Gavves
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 prototype calibration (deterministic) to the baseline multi-modal prototypes (ablation) | FewCLIP · PASCAL-5i GFSS 1-shot | mIoU_N: 33.75 -> 51.36 | Improved |
| make prototype calibration probabilistic with distribution regularization (ablation) | FewCLIP · PASCAL-5i GFSS 1-shot | hIoU: 60.79 -> 62.03 | Improved |
| make prototype calibration probabilistic (ablation) | FewCLIP · PASCAL-5i GFSS 5-shot | hIoU: 65.22 -> 66.44 | Improved |
| use calibration format P_t*P_c+P_t instead of concatenation or other operations (ablation) | FewCLIP · PASCAL-5i GFSS 1-shot | hIoU: 44.76 -> 62.03 (vs additive P_t+P_c) | Improved |
| use both images and class names for novel class registration (ablation) | FewCLIP · PASCAL-5i GFSS 1-shot | hIoU: 46.02 -> 62.03 (vs text only) | Improved |
| fine-tune only the calibration prototypes instead of visual prompts, mask decoder and background prototype (ablation) | FewCLIP · PASCAL-5i GFSS 1-shot | mIoU_B: 55.93 -> 76.68 | Improved |
| fine-tune textual prototypes on novel class samples (ablation) | FewCLIP · PASCAL-5i GFSS 1-shot | hIoU: 46.02 -> 46.83 | No clear effect |
| use FewCLIP instead of PoP on COCO-20i 5-shot | FewCLIP · COCO-20i GFSS 5-shot | hIoU: 38.77 -> 38.38 | Hurt |
| apply FewCLIP in class-incremental setting instead of PKL | PKL · PASCAL-5i CIFSS 1-shot session 5 | hIoU: 39.00 -> 61.30 | Improved |
Closest papers by meaning
- Enhancing Generalized Few-Shot Semantic Segmentation via Effective Knowledge Transfer (AAAI 2025 · similarity 0.90)
- A Surprisingly Simple Approach to Generalized Few-Shot Semantic Segmentation (NeurIPS 2024 · similarity 0.89)
- A Simple Recipe for Language-guided Domain Generalized Segmentation (CVPR 2024 · similarity 0.89)
- Toward Modality Gap: Vision Prototype Learning for Weakly-supervised Semantic Segmentation with CLIP (AAAI 2025 · similarity 0.88)
- Rethinking Prior Information Generation with CLIP for Few-Shot Segmentation (CVPR 2024 · similarity 0.88)
- Label-Efficient Few-Shot Semantic Segmentation with Unsupervised Meta-Training (AAAI 2024 · similarity 0.88)
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.
