Research map / Vision-language and multimodal / CLIP and zero-shot
Advancing Textual Prompt Learning with Anchored Attributes
Zheng Li, Yibing Song, Ming-Ming Cheng, Xiang Li, Jian Yang
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 ATPrompt attribute anchoring | CoOp · base-to-novel average, 11 datasets | harmonic mean: 71.66 -> 74.65 | Improved |
| add ATPrompt attribute anchoring | CoCoOp · base-to-novel average, 11 datasets | harmonic mean: 75.83 -> 77.95 | Improved |
| add ATPrompt attribute anchoring | MaPLe · base-to-novel average, 11 datasets | harmonic mean: 78.55 -> 79.21 | Improved |
| add ATPrompt attribute anchoring | DePT · base-to-novel average, 11 datasets | harmonic mean: 77.29 -> 78.45 | Improved |
| add ATPrompt attribute anchoring | PromptKD · base-to-novel average, 11 datasets | harmonic mean: 83.73 -> 84.35 | Improved |
| add ATPrompt attribute anchoring | CoOp · cross-dataset average | accuracy: 63.88 -> 65.26 | Improved |
| add ATPrompt attribute anchoring | MaPLe · domain generalization average | accuracy: 60.27 -> 60.60 | Improved |
| place class token at end rather than front or middle (ablation) | CoOp + ATPrompt · ImageNet | harmonic mean: 73.20 -> 73.33 | Improved |
| retain hard and soft attribute tokens in deep layers instead of dropping and re-adding (ablation) | MaPLe + ATPrompt · ImageNet | harmonic mean: 73.31 -> 73.70 | Improved |
| use searched attributes instead of irrelevant attributes (ablation) | CoOp + ATPrompt · 11-dataset average | harmonic mean: 73.59 -> 74.65 | Improved |
| use searched attributes instead of common attributes (ablation) | CoOp + ATPrompt · 11-dataset average | harmonic mean: 74.16 -> 74.65 | Improved |
| reorder the attributes (ablation) | CoOp + ATPrompt · ImageNet | harmonic mean: 73.24 vs 73.33 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- MaPLe — MaPLe: Multi-Modal Prompt Learning (CVPR 2023) · better on 4 of 4 shared results
- PromptKD — PromptKD: Unsupervised Prompt Distillation for Vision-Language Models (CVPR 2024) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Tree of Attributes Prompt Learning for Vision-Language Models (ICLR 2025 · similarity 0.92)
- Prompt Tuning In a Compact Attribute Space (AAAI 2025 · similarity 0.91)
- ArGue: Attribute-Guided Prompt Tuning for Vision-Language Models (CVPR 2024 · similarity 0.89)
- Learning to Prompt with Text Only Supervision for Vision-Language Models (AAAI 2025 · similarity 0.89)
- A Similarity Paradigm Through Textual Regularization Without Forgetting (AAAI 2025 · similarity 0.89)
- Compound Text-Guided Prompt Tuning via Image-Adaptive Cues (AAAI 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.
