# joanrod/star-vector

StarVector is a foundation model for SVG generation that transforms vectorization into a code generation task. Using a vision-language modeling architecture, StarVector processes both visual and textual inputs to produce high-quality SVG code with remarkable precision.

Repository: https://github.com/joanrod/star-vector
Canonical: https://ross.abutalabs.com/products/star-vector
Homepage: https://starvector.github.io
Language: Python
License: Apache-2.0
License Family: permissive
Topics: llm, multimodal-large-language-models, svg, vlm
Last push: 2025-11-07T20:31:24+00:00

## Health v2 (maintenance only)
Score: 49/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 51, release rhythm 35, longevity 71
- inputs: {"age_days": 996, "days_push": 299, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 4560, forks 258 (observed 2026-08-28T04:08:53.462661+00:00)

## What it is
StarVector is a foundation model that generates scalable vector graphics (SVG) code from images and text by treating vectorization as a code generation task with a vision-language architecture. It ships pretrained 1B and 8B models on HuggingFace plus the SVG-Stack dataset and SVG-Bench benchmark.

## Use cases
- convert raster images to svg code
- vectorize a png into scalable vector graphics
- generate svg icons from text descriptions
- image to svg code generation with a vision-language model
- benchmark svg generation models on svg-bench
- finetune a model for image-to-vector code generation

## When to choose
- you need high-fidelity image-to-SVG conversion with editable vector output
- you want a pretrained multimodal model for SVG code generation
- you need a research baseline or benchmark for SVG generation

## When to avoid
- you need simple one-off raster-to-vector tracing without GPU inference
- you lack GPU resources for large vision-language model inference
- you need production-grade vectorization of complex photographs

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, image-processing, llm-inference, graphics
- domain: artificial-intelligence, machine-learning, computer-vision, graphics, large-language-models
- platform: python
- tags: svg-generation, vision-language-model, image-to-code, vectorization, multimodal, foundation-model, gpu

## Member repositories
- joanrod/star-vector (main) score 49

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:53.462661+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:20:00.445550+00:00, confidence not recorded.
  - readme: https://github.com/joanrod/star-vector (fetched 2026-08-28T04:08:53.462661+00:00, sha 664614774caf)
  - homepage: https://starvector.github.io (fetched 2026-08-29T09:05:35.222164+00:00, sha 6025b0f3862a)
- Data as of 2026-08-30T08:39:29.467469+00:00.