Research map / Vision-language and multimodal / Human and embodied modeling
Don’t Forget Me: Accurate Background Recovery for Text Removal via Modeling Local-Global Context
Chongyu Liu, Lianwen Jin, Yuliang Liu, Canjie Luo, Bangdong Chen, Fengjun Guo, Kai Ding
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 HCG block (ablation) | Pix2pix baseline · SCUT-EnsText (I_out) | PSNR: 32.39 -> 32.90 | Improved |
| add HCG block (ablation) | Pix2pix baseline · SCUT-EnsText (I_out) | FID: 20.75 -> 17.40 | Improved |
| add HCG block (ablation) | Pix2pix baseline · SCUT-EnsText (I_com) | PSNR: 33.21 -> 34.88 | Improved |
| add LGCM blocks (ablation) | baseline + HCG · SCUT-EnsText (I_out) | PSNR: 32.90 -> 35.10 | Improved |
| add LGCM blocks (ablation) | baseline + HCG · SCUT-EnsText (I_out) | FID: 17.40 -> 14.36 | Improved |
| add LGCM blocks (ablation) | baseline + HCG · SCUT-EnsText (I_com) | PSNR: 34.88 -> 35.30 | Improved |
| add soft mask (ablation) | baseline + HCG + LGCM · SCUT-EnsText (I_com) | PSNR: 35.30 -> 35.83 | Improved |
| add soft mask (ablation) | baseline + HCG + LGCM · SCUT-EnsText (I_out) | PSNR: 35.10 -> 35.16 | No clear effect |
| add LCG block (ablation) | baseline + HCG + LGCM + soft mask · SCUT-EnsText (I_out) | FID: 14.33 -> 13.99 | Improved |
| add LCG block (ablation) | baseline + HCG + LGCM + soft mask · SCUT-EnsText (I_out) | PSNR: 35.16 -> 35.20 | Improved |
| add LCG block (ablation) | baseline + HCG + LGCM + soft mask · SCUT-EnsText (I_com) | FID: 15.02 -> 14.57 | Improved |
| use CTRNet instead of image inpainting methods | text removal on SCUT-EnsText · SCUT-EnsText (inpainting comparison) | PSNR: 35.41 -> 37.20 | Improved |
Closest papers by meaning
- Localized Concept Erasure for Text-to-Image Diffusion Models Using Training-Free Gated Low-Rank Adaptation (CVPR 2025 · similarity 0.83)
- ConText: Driving In-context Learning for Text Removal and Segmentation (ICML 2025 · similarity 0.82)
- Semantic Surgery: Zero-Shot Concept Erasure in Diffusion Models (NeurIPS 2025 · similarity 0.82)
- ViTEraser: Harnessing the Power of Vision Transformers for Scene Text Removal with SegMIM Pretraining (AAAI 2024 · similarity 0.82)
- Improving Image Restoration through Removing Degradations in Textual Representations (CVPR 2024 · similarity 0.82)
- TextCtrl: Diffusion-based Scene Text Editing with Prior Guidance Control (NeurIPS 2024 · similarity 0.82)
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.
