Research map / Vision-language and multimodal / CLIP and zero-shot
Open-Vocabulary Multi-Label Classification via Multi-Modal Knowledge Transfer
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use MKT instead of BiAM | Multi-label ZSL on NUS-WIDE · NUS-WIDE ZSL | mAP: 25.9 -> 37.6 | Helped |
| Use MKT instead of ZS-SDL | Multi-label GZSL on NUS-WIDE · NUS-WIDE GZSL | mAP: 12.1 -> 18.3 | Helped |
| Use MKT instead of BiAM | Multi-label ZSL on Open Images · Open Images ZSL | mAP: 65.6 -> 68.1 | Helped |
| Use MKT instead of BiAM | Multi-label GZSL on Open Images · Open Images GZSL | mAP: 81.7 -> 81.4 | Hurt |
| Add knowledge distillation (ablation) | MKT without distillation or prompt tuning · NUS-WIDE ZSL | mAP: 32.4 -> 37.3 | Helped |
| Add knowledge distillation (ablation) | MKT without distillation or prompt tuning · NUS-WIDE GZSL | mAP: 16.8 -> 18.2 | Helped |
| Add prompt tuning (ablation) | MKT without distillation or prompt tuning · NUS-WIDE ZSL | mAP: 32.4 -> 32.5 | No clear effect |
| Add prompt tuning (ablation) | MKT with distillation · NUS-WIDE ZSL | mAP: 37.3 -> 37.6 | Helped |
| Replace GloVe label embeddings with CLIP label embeddings (ablation) | MKT without distillation or prompt tuning · NUS-WIDE ZSL | mAP: 27.1 -> 32.4 | Helped |
| Replace GloVe label embeddings with CLIP label embeddings (ablation) | MKT without distillation or prompt tuning · NUS-WIDE GZSL | mAP: 16.1 -> 16.8 | Helped |
| Add the global head to the local head (ablation) | Local-head-only MKT · NUS-WIDE ZSL | mAP: 29.1 -> 32.4 | Helped |
| Add the local head to the global head (ablation) | Global-head-only MKT · NUS-WIDE ZSL | mAP: 30.3 -> 32.4 | Helped |
Closest papers by meaning
- Language-Driven Cross-Modal Classifier for Zero-Shot Multi-Label Image Recognition (ICML 2024 · similarity 0.90)
- Recover and Match: Open-Vocabulary Multi-Label Recognition through Knowledge-Constrained Optimal Transport (CVPR 2025 · similarity 0.88)
- OVMR: Open-Vocabulary Recognition with Multi-Modal References (CVPR 2024 · similarity 0.87)
- Epsilon: Exploring Comprehensive Visual-Semantic Projection for Multi-Label Zero-Shot Learning (AAAI 2025 · similarity 0.87)
- SPARC: Score Prompting and Adaptive Fusion for Zero-Shot Multi-Label Recognition in Vision-Language Models (CVPR 2025 · similarity 0.86)
- Label Propagation for Zero-shot Classification with Vision-Language Models (CVPR 2024 · similarity 0.85)
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.