SemLayer: Semantic-aware Generative Segmentation and Layer Construction for Abstract Icons

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Xu, Haiyang, Wu, Ronghuan, Wei, Li-Yi, Zhao, Nanxuan, Liu, Chenxi, Nguyen, Cuong, Tu, Zhuowen, Wang, Zhaowen
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912981652078592
author Xu, Haiyang
Wu, Ronghuan
Wei, Li-Yi
Zhao, Nanxuan
Liu, Chenxi
Nguyen, Cuong
Tu, Zhuowen
Wang, Zhaowen
author_facet Xu, Haiyang
Wu, Ronghuan
Wei, Li-Yi
Zhao, Nanxuan
Liu, Chenxi
Nguyen, Cuong
Tu, Zhuowen
Wang, Zhaowen
contents Graphic icons are a cornerstone of modern design workflows, yet they are often distributed as flattened single-path or compound-path graphics, where the original semantic layering is lost. This absence of semantic decomposition hinders downstream tasks such as editing, restyling, and animation. We formalize this problem as semantic layer construction for flattened vector art and introduce SemLayer, a visual generation empowered pipeline that restores editable layered structures. Given an abstract icon, SemLayer first generates a chromatically differentiated representation in which distinct semantic components become visually separable. To recover the complete geometry of each part, including occluded regions, we then perform a semantic completion step that reconstructs coherent object-level shapes. Finally, the recovered parts are assembled into a layered vector representation with inferred occlusion relationships. Extensive qualitative comparisons and quantitative evaluations demonstrate the effectiveness of SemLayer, enabling editing workflows previously inapplicable to flattened vector graphics and establishing semantic layer reconstruction as a practical and valuable task. Project page: https://xxuhaiyang.github.io/SemLayer/
format Preprint
id arxiv_https___arxiv_org_abs_2603_24039
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle SemLayer: Semantic-aware Generative Segmentation and Layer Construction for Abstract Icons
Xu, Haiyang
Wu, Ronghuan
Wei, Li-Yi
Zhao, Nanxuan
Liu, Chenxi
Nguyen, Cuong
Tu, Zhuowen
Wang, Zhaowen
Computer Vision and Pattern Recognition
Graphics
Human-Computer Interaction
Graphic icons are a cornerstone of modern design workflows, yet they are often distributed as flattened single-path or compound-path graphics, where the original semantic layering is lost. This absence of semantic decomposition hinders downstream tasks such as editing, restyling, and animation. We formalize this problem as semantic layer construction for flattened vector art and introduce SemLayer, a visual generation empowered pipeline that restores editable layered structures. Given an abstract icon, SemLayer first generates a chromatically differentiated representation in which distinct semantic components become visually separable. To recover the complete geometry of each part, including occluded regions, we then perform a semantic completion step that reconstructs coherent object-level shapes. Finally, the recovered parts are assembled into a layered vector representation with inferred occlusion relationships. Extensive qualitative comparisons and quantitative evaluations demonstrate the effectiveness of SemLayer, enabling editing workflows previously inapplicable to flattened vector graphics and establishing semantic layer reconstruction as a practical and valuable task. Project page: https://xxuhaiyang.github.io/SemLayer/
title SemLayer: Semantic-aware Generative Segmentation and Layer Construction for Abstract Icons
topic Computer Vision and Pattern Recognition
Graphics
Human-Computer Interaction
url https://arxiv.org/abs/2603.24039