OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Developer Tools & SDKs

Glue Factory

Glue Factory is a Python library for training and evaluating deep neural networks that detect and match local visual features, including image points and lines.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Glue Factory — OpenViST project profile in AI Developer Tools & SDKs.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Glue Factory when you need a training and evaluation workflow for local image-feature detection and matching, rather than only a prebuilt application for viewing or editing images.

Who it is for

  • Computer-vision researchers reproducing point- or line-matching model training.
  • Developers training matchers with multiple datasets and their own local features or lines.
  • Teams comparing feature extractors or matchers on established benchmarks.

Documented uses

  • Reproduce training workflows for models such as LightGlue and GlueStick.
  • Train visual feature models with multiple datasets and custom feature inputs.
  • Evaluate feature extractors or matchers on HPatches and MegaDepth-1500.
Before you choose

Requirements & limits

The repository is public and the code and trained models are released under Apache-2.0. Installation requires Python 3 and PyTorch.

  • The library is built around Python 3 and PyTorch.
  • Basic dependencies can be installed in editable mode; some advanced features may require the extra dependency set.
  • Benchmark evaluations download datasets automatically and require local disk space.
  • Some third-party models are distributed separately under potentially restrictive licenses.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Evaluate LightGlue on HPatches

Run the documented HPatches evaluation for the pretrained SuperPoint+LightGlue model.

Before you start

  • Python 3 and PyTorch
  • Glue Factory installed
  • About 1.8 GB of free disk space for the automatically downloaded HPatches dataset

Steps

  1. Install Glue Factory from its repository using the documented editable pip installation.

  2. Run the command below.

    Documented command
    python -m gluefactory.eval.hpatches --conf superpoint+lightglue-official --overwrite

Expected outcome

The evaluation reports matching and homography-error metrics for the pretrained SuperPoint+LightGlue model.

Guide limitations

  • The dataset downloads automatically, by default to the data/ directory.
  • The documented command uses the default OpenCV robust estimator; the README separately shows a PoseLib option.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
14 October 2026
Original website title
cvg/glue-factory · Official homepage title · retained in its source language
How we review sourcesSuggest a correction