OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Gluestick

GlueStick is a deep-learning image matcher that jointly finds corresponding points and lines in two images. Its public repository contains the implementation and pretrained weights, along with a Python command-line inference path; training code is provided separately through GlueFactory.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Gluestick — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use GlueStick when an image-pair workflow needs correspondences that include both points and lines, and you can prepare the documented local Python environment.

Who it is for

  • Computer-vision practitioners matching image pairs for visual localization.
  • Teams preparing image correspondences for structure-from-motion or mapping workflows.
  • Researchers who need a documented starting point for running pretrained point-and-line matching inference.

Documented uses

  • Find point and line correspondences between two images.
  • Run inference with the provided pretrained checkpoint from a Python command.
  • Use matched image features as input to a localization, mapping, or structure-from-motion workflow.
  • Explore training through the separate GlueFactory framework.
Before you choose

Requirements & limits

The repository is public and describes its code as MIT-licensed. Its README cautions that the SuperPoint backbone has a non-commercial license and describes the overall system as non-commercial.

  • The documented setup targets Ubuntu 22.04 and requires build tools and OpenCV development packages.
  • Inference requires downloading the pretrained checkpoint and providing two image paths.
  • Training is handled in the separate GlueFactory repository.
  • The MIT code license does not remove the README's stated non-commercial constraint associated with the SuperPoint backbone.

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

Install GlueStick and run inference on an image pair

Set up the documented Ubuntu environment, obtain pretrained weights, and invoke inference with two image paths.

Before you start

  • Ubuntu 22.04, as specified by the README
  • Build tools, CMake, and OpenCV development packages
  • Python and a virtual environment
  • Two image files and the pretrained checkpoint

Steps

  1. Install build-essential, cmake, libopencv-dev, and libopencv-contrib-dev with apt-get.

  2. Clone the repository recursively, enter its directory, create and activate a Python virtual environment, then install requirements and the package.

  3. Download checkpoint_GlueStick_MD.tar to resources/weights using the documented wget command.

  4. Run python -m gluestick.run with -img1 and -img2 set to the two image paths.

Expected outcome

The documented command runs GlueStick inference for the supplied image pair using the downloaded pretrained weights.

Guide limitations

  • The README documents the command and setup, but does not specify an output format or downstream pipeline integration.
  • The README says the overall system uses a SuperPoint backbone with a non-commercial license.
Documentation 1 ↗
Evidence and transparency

Sources & review

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