Lists
Lists
Chavan, Rutuja, Hirve, Sahil, Shinde, Swati, Virdee, Bal Singh and Khanna, Ashish (2025) Enhancing text-specific inpainting using diffusion models and OCR. In: 2025 9th International Conference on Computing, Communication, Control and Automation (ICCCBEA), 22-23 August 2025, Pune, India.
Text-specific inpainting-masking and replacing particular words and phrases in the different images. It has significant potential for applications such as document redaction, privacy protection, and automated image editing. While previous research has struggled with this task. This paper presents an approach combining Optical Character Recognition (OCR) and diffusion models to address these challenges. We use pytesseract for text detection and Stable Diffusion for inpainting, after using that we aim to accurately replace specific words in images. Our experiments show promising results in simpler cases but reveal limitations when handling intricate backgrounds and fonts. Based on these findings, we suggest improvements to enhance the robustness of the method, specifically in handling complex image environments.
Available under License Creative Commons Attribution 4.0.
Download (522kB) | Preview
 |
View Item |