https://github.com/kingnobro/layerpeeler

(Siggraph Asia 2025) Code of "LayerPeeler: Autoregressive Peeling for Layer-wise Image Vectorization"
https://github.com/kingnobro/layerpeeler

image-vectorization layered-layout vectorization

Last synced: 27 days ago
JSON representation

(Siggraph Asia 2025) Code of "LayerPeeler: Autoregressive Peeling for Layer-wise Image Vectorization"

• Host: GitHub
• URL: https://github.com/kingnobro/layerpeeler
• Owner: kingnobro
• Created: 2025-06-20T04:01:13.000Z (4 months ago)
• Default Branch: main
• Last Pushed: 2025-09-06T15:49:07.000Z (about 1 month ago)
• Last Synced: 2025-09-06T17:39:15.611Z (about 1 month ago)
• Topics: image-vectorization, layered-layout, vectorization
• Homepage: https://layerpeeler.github.io/
• Size: 1.21 MB
• Stars: 0
• Watchers: 0
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# LayerPeeler: Autoregressive Peeling for Layer-wise Image Vectorization

[![arXiv](https://img.shields.io/badge/arXiv-2312.16476-b31b1b.svg)](https://arxiv.org/abs/2505.23740)

[![website](https://img.shields.io/badge/Website-Gitpage-4CCD99)](https://layerpeeler.github.io/)

![title](./assets/teaser.png)

## Overview

LayerPeeler is a framework for layer-wise image vectorization that decomposes images into structured vector representations. The system uses a vision-language model to analyze the image and construct a hierarchical layer graph, identifying the topmost visual elements. These detected elements are then processed by a fine-tuned image diffusion model to generate clean background images with the specified elements removed, enabling precise layer-by-layer vectorization.

## Updates

- [ ] Dataset Released

- [ ] Training Code Released

- [ ] Layer Graph Generation Released

- [ ] Image Diffusion Model Released

- [ ] Vectorization Tools Released