Learning Feature-Preserving Portrait Editing from Generated Pairs

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Chen, Bowei, Zhi, Tiancheng, Zhu, Peihao, Sang, Shen, Liu, Jing, Luo, Linjie
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910546079514624
author Chen, Bowei
Zhi, Tiancheng
Zhu, Peihao
Sang, Shen
Liu, Jing
Luo, Linjie
author_facet Chen, Bowei
Zhi, Tiancheng
Zhu, Peihao
Sang, Shen
Liu, Jing
Luo, Linjie
contents Portrait editing is challenging for existing techniques due to difficulties in preserving subject features like identity. In this paper, we propose a training-based method leveraging auto-generated paired data to learn desired editing while ensuring the preservation of unchanged subject features. Specifically, we design a data generation process to create reasonably good training pairs for desired editing at low cost. Based on these pairs, we introduce a Multi-Conditioned Diffusion Model to effectively learn the editing direction and preserve subject features. During inference, our model produces accurate editing mask that can guide the inference process to further preserve detailed subject features. Experiments on costume editing and cartoon expression editing show that our method achieves state-of-the-art quality, quantitatively and qualitatively.
format Preprint
id arxiv_https___arxiv_org_abs_2407_20455
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Learning Feature-Preserving Portrait Editing from Generated Pairs
Chen, Bowei
Zhi, Tiancheng
Zhu, Peihao
Sang, Shen
Liu, Jing
Luo, Linjie
Computer Vision and Pattern Recognition
Portrait editing is challenging for existing techniques due to difficulties in preserving subject features like identity. In this paper, we propose a training-based method leveraging auto-generated paired data to learn desired editing while ensuring the preservation of unchanged subject features. Specifically, we design a data generation process to create reasonably good training pairs for desired editing at low cost. Based on these pairs, we introduce a Multi-Conditioned Diffusion Model to effectively learn the editing direction and preserve subject features. During inference, our model produces accurate editing mask that can guide the inference process to further preserve detailed subject features. Experiments on costume editing and cartoon expression editing show that our method achieves state-of-the-art quality, quantitatively and qualitatively.
title Learning Feature-Preserving Portrait Editing from Generated Pairs
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2407.20455