@inproceedings{b2d099d98ee64b45b541b2342ef8005a,
title = "MaskDiffuse: Text-Guided Face Mask Removal Based on Diffusion Models",
abstract = "As masked face images can significantly degrade the performance of face-related tasks, face mask removal remains an important and challenging task. In this paper, we propose a novel learning framework, called MaskDiffuse, to remove face masks based on Denoising Diffusion Probabilistic Model (DDPM). In particular, we leverage CLIP to fill the missing parts by guiding the reverse process of pretrained diffusion model with text prompts. Furthermore, we propose a multi-stage blending strategy to preserve the unmasked areas and a conditional resampling approach to make the generated contents consistent with the unmasked regions. Thus, our method achieves interactive user-controllable and identity-preserving masking removal with high quality. Both qualitative and quantitative experimental results demonstrate the superiority of our method for mask removal over alternative methods.",
keywords = "Diffusion models, Mask removal, Text-to-image",
author = "Jingxia Lu and Xianxu Hou and Hao Li and Zhibin Peng and Linlin Shen and Lixin Fan",
note = "Publisher Copyright: {\textcopyright} 2024, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.; 6th Chinese Conference on Pattern Recognition and Computer Vision, PRCV 2023 ; Conference date: 13-10-2023 Through 15-10-2023",
year = "2024",
doi = "10.1007/978-981-99-8537-1_35",
language = "English",
isbn = "9789819985364",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Science and Business Media Deutschland GmbH",
pages = "435--446",
editor = "Qingshan Liu and Hanzi Wang and Rongrong Ji and Zhanyu Ma and Weishi Zheng and Hongbin Zha and Xilin Chen and Liang Wang",
booktitle = "Pattern Recognition and Computer Vision - 6th Chinese Conference, PRCV 2023, Proceedings",
}