Research map / Image generation / Editing and style transfer
CLIP-FTI: Fine-Grained Face Template Inversion via CLIP-Driven Attribute Conditioning
Longchen Dai, Zixuan Shen, Zhiheng Zhou, Peipeng Yu, Zhihua Xia
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 CLIP attribute embeddings (ablation) | CLIP-FTI · LFW | Type-II TAR: 0.4655 -> 0.8174 | Improved |
| add CLIP attribute embeddings (ablation) | CLIP-FTI · LFW | FAMSE: 0.0503 -> 0.0451 | Improved |
| use multi-head attention fusion rather than concatenation (ablation) | CLIP-FTI · LFW | Type-II TAR: 0.4712 -> 0.8174 | Improved |
| use template as the attention query (ablation) | CLIP-FTI · LFW | Type-II TAR: 0.4792 -> 0.8174 | Improved |
| add attribute loss Lattr alongside LPIPS loss (ablation) | CLIP-FTI · LFW | Type-II TAR: 0.6447 -> 0.7229 | Improved |
| add attribute loss Lattr alongside LPIPS loss (ablation) | CLIP-FTI · LFW | FAMSE: 0.0428 -> 0.0451 | Hurt |
| add LPIPS loss alongside attribute loss (ablation) | CLIP-FTI · LFW | Type-II TAR: 0.4453 -> 0.7229 | Improved |
| add attribute loss Lattr alongside LPIPS loss (ablation) | CLIP-FTI · LFW | Type-I TAR: 0.9819 -> 0.9822 | No clear effect |
| use CLIP-FTI instead of Otroshi et al. inversion (ablation) | CLIP-FTI · LFW (db ArcFace) | Type-II TAR@FAR=0.1%: 0.4655 -> 0.8174 | Improved |
| use CLIP-FTI instead of Otroshi et al. inversion | CLIP-FTI · LFW (db ElasticFace, surrogate ArcFace) | Type-I TAR@FAR=0.1%: 0.9486 -> 0.9435 | Hurt |
| use CLIP-FTI instead of Otroshi et al. for cross-architecture transfer | CLIP-FTI · LFW to HRNet | Type-I TAR@FAR=0.1% (%): 51.63 -> 65.23 | Improved |
Closest papers by meaning
- On the Reconstruction of Face Images from Deep Face Templates (TPAMI 2018 · similarity 0.86)
- Face Reconstruction from Facial Templates by Learning Latent Space of a Generator Network (NeurIPS 2023 · similarity 0.86)
- Template Inversion Attack against Face Recognition Systems using 3D Face Reconstruction (ICCV 2023 · similarity 0.86)
- StyleIPSB: Identity-Preserving Semantic Basis of StyleGAN for High Fidelity Face Swapping (CVPR 2023 · similarity 0.86)
- FaceA-Net: Facial Attribute-Driven ID Preserving Image Generation Network (AAAI 2025 · similarity 0.85)
- When StyleGAN Meets Stable Diffusion: a W+ Adapter for Personalized Image Generation (CVPR 2024 · similarity 0.84)
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.
