Human Body Restoration with One-Step Diffusion Model and A New Benchmark

Jue Gong, Jingkai Wang, Zheng Chen, Xing Liu, Hong Gu, Yulun Zhang, Xiaokang Yang
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:20016-20026, 2025.

Abstract

Human body restoration, as a specific application of image restoration, is widely applied in practice and plays a vital role across diverse fields. However, thorough research remains difficult, particularly due to the lack of benchmark datasets. In this study, we propose a high-quality dataset automated cropping and filtering (HQ-ACF) pipeline. This pipeline leverages existing object detection datasets and other unlabeled images to automatically crop and filter high-quality human images. Using this pipeline, we constructed a person-based restoration with sophisticated objects and natural activities (PERSONA) dataset, which includes training, validation, and test sets. The dataset significantly surpasses other human-related datasets in both quality and content richness. Finally, we propose OSDHuman, a novel one-step diffusion model for human body restoration. Specifically, we propose a high-fidelity image embedder (HFIE) as the prompt generator to better guide the model with low-quality human image information, effectively avoiding misleading prompts. Experimental results show that OSDHuman outperforms existing methods in both visual quality and quantitative metrics. The dataset and code are available at: https://github.com/gobunu/OSDHuman.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-gong25f, title = {Human Body Restoration with One-Step Diffusion Model and A New Benchmark}, author = {Gong, Jue and Wang, Jingkai and Chen, Zheng and Liu, Xing and Gu, Hong and Zhang, Yulun and Yang, Xiaokang}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {20016--20026}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/gong25f/gong25f.pdf}, url = {https://proceedings.mlr.press/v267/gong25f.html}, abstract = {Human body restoration, as a specific application of image restoration, is widely applied in practice and plays a vital role across diverse fields. However, thorough research remains difficult, particularly due to the lack of benchmark datasets. In this study, we propose a high-quality dataset automated cropping and filtering (HQ-ACF) pipeline. This pipeline leverages existing object detection datasets and other unlabeled images to automatically crop and filter high-quality human images. Using this pipeline, we constructed a person-based restoration with sophisticated objects and natural activities (PERSONA) dataset, which includes training, validation, and test sets. The dataset significantly surpasses other human-related datasets in both quality and content richness. Finally, we propose OSDHuman, a novel one-step diffusion model for human body restoration. Specifically, we propose a high-fidelity image embedder (HFIE) as the prompt generator to better guide the model with low-quality human image information, effectively avoiding misleading prompts. Experimental results show that OSDHuman outperforms existing methods in both visual quality and quantitative metrics. The dataset and code are available at: https://github.com/gobunu/OSDHuman.} }
Endnote
%0 Conference Paper %T Human Body Restoration with One-Step Diffusion Model and A New Benchmark %A Jue Gong %A Jingkai Wang %A Zheng Chen %A Xing Liu %A Hong Gu %A Yulun Zhang %A Xiaokang Yang %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-gong25f %I PMLR %P 20016--20026 %U https://proceedings.mlr.press/v267/gong25f.html %V 267 %X Human body restoration, as a specific application of image restoration, is widely applied in practice and plays a vital role across diverse fields. However, thorough research remains difficult, particularly due to the lack of benchmark datasets. In this study, we propose a high-quality dataset automated cropping and filtering (HQ-ACF) pipeline. This pipeline leverages existing object detection datasets and other unlabeled images to automatically crop and filter high-quality human images. Using this pipeline, we constructed a person-based restoration with sophisticated objects and natural activities (PERSONA) dataset, which includes training, validation, and test sets. The dataset significantly surpasses other human-related datasets in both quality and content richness. Finally, we propose OSDHuman, a novel one-step diffusion model for human body restoration. Specifically, we propose a high-fidelity image embedder (HFIE) as the prompt generator to better guide the model with low-quality human image information, effectively avoiding misleading prompts. Experimental results show that OSDHuman outperforms existing methods in both visual quality and quantitative metrics. The dataset and code are available at: https://github.com/gobunu/OSDHuman.
APA
Gong, J., Wang, J., Chen, Z., Liu, X., Gu, H., Zhang, Y. & Yang, X.. (2025). Human Body Restoration with One-Step Diffusion Model and A New Benchmark. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:20016-20026 Available from https://proceedings.mlr.press/v267/gong25f.html.

Related Material