Research map / Vision-language and multimodal / CLIP and zero-shot
Plug-and-play Class-aware Knowledge Injection for Prompt Learning with Visual-Language Model
Junhui Yin, Nan Pu, Xinyu Zhang, Lingfeng Yang, Lin Yuanbo Wu, Xiaojie Wang, Zhun Zhong
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 CAKI class-aware knowledge injection | CoOp · Flower102 base-to-novel (1-shot) | harmonic mean accuracy: 74.1 -> 77.4 | Improved |
| add CAKI | PromptSRC · Aircraft base-to-novel (1-shot) | harmonic mean accuracy: 10.7 -> 23.5 | Improved |
| add CAKI | GalLoP · DTD base-to-novel (16-shot) | harmonic mean accuracy: 68.7 -> 69.3 | Improved |
| add CAKI | PromptSRC · ImageNet few-shot (4-shot) | accuracy: 73.2 -> 74.1 | Improved |
| add CAKI | CLIP-LoRA · ImageNet few-shot (4-shot) | accuracy: 71.3 -> 72.6 | Improved |
| add CAKI | L2C · DomainNet (average over domains) | accuracy: 61.2 -> 62.4 | Improved |
| add CAKI | CoCoOp (SEEM-Tiny) · Semantic segmentation ADE | mIoU: 20.2 -> 21.6 | Improved |
| add query-key prompt matching to class-specific prompt generation (ablation) | CAKI (CoOp baseline) · Flower102 base-to-novel (1-shot) | harmonic mean accuracy: 75.8 -> 77.4 | Improved |
| add class-specific prompt generation to query-key matching (ablation) | CAKI (CoOp baseline) · Pets base-to-novel (1-shot) | harmonic mean accuracy: 95.4 -> 96.1 | Improved |
| use query-key matching instead of random prompt ensemble (Prompt-R) (ablation) | CAKI (CoOp baseline) · DTD base-to-novel (1-shot) | accuracy: 56.2 -> 57.2 | Improved |
| use query-key matching instead of ensembling all prompts (Prompt-A) (ablation) | CAKI (CoOp baseline) · DTD base-to-novel (1-shot) | accuracy: 55.9 -> 57.2 | Improved |
| add instance-specific prompts to class-shared prompts (ablation) | CAKI prompt combination · Flower102 base-to-novel (1-shot) | harmonic mean accuracy: 74.0 -> 71.4 | Hurt |
Closest papers by meaning
- PromptKD: Unsupervised Prompt Distillation for Vision-Language Models (CVPR 2024 · similarity 0.92)
- Learning to Prompt with Text Only Supervision for Vision-Language Models (AAAI 2025 · similarity 0.91)
- ProAPO: Progressively Automatic Prompt Optimization for Visual Classification (CVPR 2025 · similarity 0.90)
- Learning to Learn Better Visual Prompts (AAAI 2024 · similarity 0.90)
- Prompt Tuning In a Compact Attribute Space (AAAI 2025 · similarity 0.90)
- SEC-Prompt:SEmantic Complementary Prompting for Few-Shot Class-Incremental Learning (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.
