mlsanityData to 14 Oct 2026

What Shape Is Optimal for Masks in Text Removal?

arXiv 2511.22499 ↗First posted 27 Nov 2025cs.CV cs.CL cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

The advent of generative models has dramatically improved the accuracy of image inpainting. In particular, by removing specific text from document images, reconstructing original images is extremely important for industrial applications. However, most existing methods of text removal focus on deleting simple scene text which appears in images captured by a camera in an outdoor environment. There is little research dedicated to complex and practical images with dense text. Therefore, we created benchmark data for text removal from images including a large amount of text. From the data, we found that text-removal performance becomes vulnerable against mask profile perturbation. Thus, for practical text-removal tasks, precise tuning of the mask shape is essential. This study developed a method to model highly flexible mask profiles and learn their parameters using Bayesian optimization. The resulting profiles were found to be character-wise masks. It was also found that the minimum cover of a text region is not optimal. Our research is expected to pave the way for a user-friendly guideline for manual masking.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 23 of its 48 references that are in the map. First posted on arXiv 27 Nov 2025.

Builds on

23 of 48 references.

References · 23

  1. 2019Optuna: A Next-generation Hyperparameter Optimization Frameworkcited by 2,501
  2. 2022RePaint: Inpainting using Denoising Diffusion Probabilistic Modelscited by 1,430
  3. 2021Real-ESRGAN: Training Real-World Blind Super-Resolution with Pure Synthetic Datacited by 970
  4. 2018Generative Image Inpainting with Contextual Attentioncited by 580
  5. 2018Image Inpainting for Irregular Holes Using Partial Convolutionscited by 505
  6. 2023Rethinking FID: Towards a Better Evaluation Metric for Image Generationcited by 310
  7. 2024BrushNet: A Plug-and-Play Image Inpainting Model with Decomposed Dual-Branch Diffusioncited by 202
  8. 2024OCRBench v2: An Improved Benchmark for Evaluating Large Multimodal Models on Visual Text Localization and Reasoningcited by 173
  9. 2019Image inpainting: A reviewcited by 89
  10. 2023Review of Large Vision Models and Visual Prompt Engineeringcited by 77
  11. 2022Recent Advances in Bayesian Optimizationcited by 76
  12. 2024Visual Prompting in Multimodal Large Language Models: A Surveycited by 54
11 more
  1. 2019MTRNet: A Generic Scene Text Erasercited by 29
  2. 2018EnsNet: Ensconce Text in the Wildcited by 27
  3. 2017Scene Text Erasercited by 26
  4. 2019MTRNet++: One-stage Mask-based Scene Text Erasercited by 24
  5. 2021Stroke-Based Scene Text Erasing Using Synthetic Data for Trainingcited by 20
  6. 2020Scene text removal via cascaded text stroke detection and erasingcited by 18
  7. 2022Don't Forget Me: Accurate Background Recovery for Text Removal via Modeling Local-Global Contextcited by 18
  8. 2023FETNet: Feature Erasing and Transferring Network for Scene Text Removalcited by 12
  9. 2025syftr: Pareto-Optimal Generative AIcited by 5
  10. 2024DiffSTR: Controlled Diffusion Models for Scene Text Removalcited by 3
  11. 2025Robustness of Structured Data Extraction from Perspectively Distorted Documentscited by 1

Built on by

Not cited yet.