# google-research/inksight

Repository: https://github.com/google-research/inksight
Canonical: https://ross.abutalabs.com/products/inksight
Language: Jupyter Notebook
License: Apache-2.0
License Family: permissive
Last push: 2026-08-09T21:18:34+00:00

## Health v2 (maintenance only)
Score: 65/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 96, release rhythm 35, longevity 48
- inputs: {"age_days": 679, "days_push": 24, "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 1006, forks 74 (observed 2026-08-28T04:03:11.707414+00:00)

## What it is
InkSight is a Google Research system that converts photos of offline handwritten text into digital ink strokes using a ViT and mT5 encoder-decoder vision-language model. The repository provides research code, notebooks, and model outputs for offline-to-online handwriting conversion.

## Use cases
- convert photos of handwritten notes into digital ink
- derender handwriting images into stroke sequences
- digitize handwritten text from paper
- run handwriting-to-vector conversion in a notebook
- experiment with vision-language models for handwriting

## When to choose
- you need to transform handwritten note photos into editable digital ink
- you want to reproduce or build on the InkSight research models
- you need handwriting derendering with a pretrained model

## When to avoid
- you need production-grade OCR of printed text
- you want a polished end-user application rather than research code
- you need handwriting recognition without stroke-level output

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, computer-vision, nlp, image-processing
- domain: machine-learning, computer-vision
- platform: python
- tags: handwriting-recognition, digital-ink, vision-language-models, derendering, research-code, natural-language-processing, gpu

## Member repositories
- google-research/inksight (main) score 65

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:11.707414+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 2026-08-30T07:12:46.566634+00:00, confidence not recorded.
  - readme: https://github.com/google-research/inksight (fetched 2026-08-28T04:03:11.707414+00:00, sha 0418b8ce72b6)
- Data as of 2026-08-30T08:39:29.467469+00:00.