Research map / Vision-language and multimodal / CLIP and zero-shot
CLIP2Pose: Frozen CLIP as Semantic Guide for Domain Adaptive Pose Estimation
Jiawen Li, Fei Jiang, Dandan Zhu, Jinxin Shi, Aimin Zhou
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 Semantic-Driven Prompt (SDP) (ablation) | CLIP2Pose · SURREAL->Human3.6M | PCK@0.05 (All): 75.2 -> 77.1 | Improved |
| add Semantic Modulation of Visual Features (SMVF) on top of SDP (ablation) | CLIP2Pose · SURREAL->Human3.6M | PCK@0.05 (All): 77.1 -> 78.7 | Improved |
| add Directional Alignment Loss (DAL) (ablation) | CLIP2Pose · SURREAL->Human3.6M | PCK@0.05 (All): 78.7 -> 79.9 | Improved |
| add Semantic-Driven Prompt (SDP) (ablation) | CLIP2Pose · RHD->H3D | PCK@0.05 (All): 75.8 -> 78.9 | Improved |
| add SMVF on top of SDP (ablation) | CLIP2Pose · RHD->H3D | PCK@0.05 (All): 78.9 -> 81.5 | Improved |
| add DAL on top of SDP and SMVF (ablation) | CLIP2Pose · RHD->H3D | PCK@0.05 (All): 81.5 -> 83.6 | Improved |
| use SDP instead of CoOp / CoCoOp prompts (ablation) | CLIP2Pose · SURREAL->Human3.6M and RHD->H3D | PCK@0.05 | Improved |
| vary GCN layers, lambda, token length around optimum (n=3, lambda=0.3, m=10) (ablation) | CLIP2Pose · RHD->H3D and SURREAL->Human3.6M | PCK@0.05 | No clear effect |
| use frozen CLIP prompt guidance vs prior CLIP-based UDA (AD-CLIP) | AD-CLIP · SURREAL->Human3.6M | PCK@0.05 (All): 75.4 -> 79.9 | Improved |
| use CLIP2Pose vs UniFrame in synthetic-to-wild | UniFrame · RHD->WBH | PCK@0.05 (All): 26.9 -> 31.2 | Improved |
| use CLIP2Pose on unseen target domain | UniFrame · Train RHD->H3D, test FreiHand | PCK@0.05 (All): 47.1 -> 55.2 | Improved |
| use CLIP2Pose vs best prior on LSP | DA-LLPose · SURREAL->LSP | PCK@0.05 (All): 83.5 -> 84.2 | Improved |
Closest papers by meaning
- SmartCLIP: Modular Vision-language Alignment with Identification Guarantees (CVPR 2025 · similarity 0.88)
- Harnessing Frozen Unimodal Encoders for Flexible Multimodal Alignment (CVPR 2025 · similarity 0.88)
- Domain-Agnostic Mutual Prompting for Unsupervised Domain Adaptation (CVPR 2024 · similarity 0.87)
- Domain Generalization in CLIP via Learning with Diverse Text Prompts (CVPR 2025 · similarity 0.87)
- UniPose: A Unified Multimodal Framework for Human Pose Comprehension, Generation and Editing (CVPR 2025 · similarity 0.87)
- Rethinking Misalignment in Vision-Language Model Adaptation from a Causal Perspective (NeurIPS 2024 · similarity 0.87)
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.
