deep-image-matching

DEEP-IMAGE-MATCHING

SIFT	DISK	IMAGES ORIENTATION	DENSE WITH ROMA

SIFT	SUPERGLUE

Multivew matcher for SfM software. Support both deep-learning based and hand-crafted local features and matchers and export keypoints and matches directly in a COLMAP database or to Agisoft Metashape by importing the reconstruction in Bundler format. Now, it supports both OpenMVG and MicMac. Feel free to collaborate!

For how to use DIM, check the Documentation (updated for the master branch).

Please, note that deep-image-matching is under active development and it is still in an experimental stage. If you find any bug, please open an issue. For the licence of individual local features and matchers please refer to the authors' original projects.

Key features:

• Multiview
• Large format images
• SOTA deep-learning and hand-crafted features
• Support for image rotations
• Compatibility with several SfM software
• Support image retrieval with deep-learning local features
• Graph-based clustering
• Run SfM directly in DIM (pycolmap, openmvg, etc)

Supported Extractors

Algorithm	Year	Paper link	Github link	Notes
RIPE	2025	link	link	supported
RDD sparse	2025	link	link	supported
LiftFeat	2025	link	link	supported
XFeat	2024	link	link	supported
DeDoDe	2024	link	link	only GPU
ALIKED	2023	link	link	supported
SRIF	2023	link	link	supported
DISK	2020	link	link	supported
KeyNet	2019	link	link	supported
SuperPoint	2018	link	link	supported
Superpoint open	2018	link	link	supported
HardNet	2017	link	link	supported
ORB	2011	link	link	from OpenCV
SIFT	2004	link	link	from OpenCV

Supported Matchers

Algorithm	Year	Paper link	Github link	Notes
LightGlue	2023	link	link	with SuperPoint, DISK, and ALIKED
LighterGlue	2023	link	link	with XFeat
RoMa	2023	link	link	supported
SE2-LoFTR	2022	link	link	no tiling and only GPU
LoFTR	2021	link	link	only GPU
SuperGlue	2020	link	link	with SuperPoint
Nearest Neighbor	-	-	-	from KORNIA

Supported SfM software

| ✓ COLMAP | | ✓ OpenMVG | | ✓ MICMAC | | ✓ Agisoft Metashape | | ✓ Software that supports bundler format |

Colab demo and notebooks

Want to run on a sample dataset? ➡️

Want to run on your images? ➡️

DIM can also be utilized as a library instead of being executed through the Command Line Interface (refer to the Usage Instructions).

For quick examples, see:

• demo.py - Simple script demonstrating the basic workflow
• demo.ipynb - Interactive notebook version of the demo
• notebooks/sfm_pipeline.ipynb - Complete SfM pipeline with detailed explanations

Local Installation

For installing deep-image-matching, we recommend using uv for fast and reliable package management:

# Install uv if you haven't already
curl -LsSf https://astral.sh/uv/install.sh | sh

# Create and activate a virtual environment
uv venv --python 3.9
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

Then, you can install deep-image-matching using uv:

uv pip install -e .

This command will install the package in editable mode, allowing you to modify the source code and see changes immediately without needing to reinstall. If you want to use deep-image-matching as a non-editable library, you can also install it without the -e flag.

This will also install pycolmap as a dependency, which is required for running the 3D reconstruction. If you have any issues with pycolmap, you can manually install it following the official instructions here.

To verify that deep-image-matching is correctly installed, you can try to import the package in a Python shell:

import deep_image_matching as dim

To test most of the functionality, run the tests to check if deep-image-matching is correctly installed, run:

uv run pytests

For more information, check the documentation.

Why uv?

This project has migrated from conda/pip to uv for dependency management. Benefits include:

• Faster installation: uv is significantly faster than pip for dependency resolution and installation
• Better dependency resolution: More reliable resolution of complex dependency trees
• Lockfile support: uv.lock ensures reproducible installations across different environments
• Integrated tooling: Built-in support for virtual environments, Python version management, and project building
• Cross-platform consistency: Better support for different operating systems and architectures

Conda/pip installation

If you have any issue with uv, you prefer to have a global installation of DIM, or you have any other problem with the installation, you can use conda/manba to create an environment and install DIM from source using pip:

git clone https://github.com/3DOM-FBK/deep-image-matching.git
cd deep-image-matching

conda create -n deep-image-matching python=3.9
conda activate deep-image-matching
pip install -e .

Docker Installation

For Docker installation, see the Docker Installation section in the documentation.

Usage instructions

Quick start with the demo

For a quick start, check out the demo.py script or demo.ipynb notebook that demonstrate basic usage with the example dataset:

python demo.py --dir assets/example_cyprus --pipeline superpoint+lightglue

The demo runs the complete pipeline from feature extraction to 3D reconstruction using the provided example dataset.

A similar demo example is also available as a notebook in demo.ipynb.

Command Line Interface

Use the following command to see all the available options from the CLI:

python -m deep_image_matching --help

For example, to run the matching with SuperPoint and LightGlue on the example_cyprus dataset:

python -m deep_image_matching --dir assets/example_cyprus --pipeline superpoint+lightglue

The --dir parameter defines the processing directory, where all the results will be saved. This directory must contain a subfolder named images with all the images to be processed.

Library usage

Deep-image-matching can also be used as a Python library. For a comprehensive example showing the complete SfM pipeline, see notebooks/sfm_pipeline.ipynb.

For detailed usage instructions and configurations, refer to the documentation.

Advanced usage

For advanced usage, please refer to the documentation and/or check the scripts directory.

Merging databases with different local features

To run the matching with different local features and/or matchers and marging together the results, you can use scripts in the ./scripts directory for merging the COLMAP databases.

python ./join_databases.py --help
python ./join_databases.py --input path/to/dir/with/databases --output path/to/output/dir

Exporting the solution to Metashape

Suggested solution:

• It is now possible to run SfM directly in Metashape using 2D observations extracted in DIM. You can use the script export_to_bundler.py from the scripts folder. It will create a fake bundler file. Then in Metashape import all the images you need, import camera poses using the bundler file, select all images and reset the alignment. Finally right click, align selected cameras (see issue).

Other solutions:

• To export the solution to Metashape, you can export the COLMAP database to Bundler format and then import it into Metashape. This can be done from Metashape GUI, by first importing the images and then use the function Import Cameras (File -> Import -> Import Cameras) to select Bundler file (e.g., bundler.out) and the image list file (e.g., bundler_list.txt).

• Alternatevely, you can use the export_to_metashape.py script to automatically create a Metashape project from a reconstruction saved in Bundler format. The script export_to_metashape.py takes as input the solution in Bundler format and the images and it exports the solution to Metashape. It requires to install Metashape as a Python module in your environment and to have a valid license. Please, refer to the instructions at https://github.com/franioli/metashape.

How to contribute

Any contribution to this repo is really welcome! If you want to contribute to the project, please, check the contributing guidelines.

To Do List

See the TODO list for the list of features and improvements that are planned for the future.

References

If you find the repository useful for your work consider citing the papers:

@article{morelli2024_deep_image_matching,
  AUTHOR = {Morelli, L. and Ioli, F. and Maiwald, F. and Mazzacca, G. and Menna, F. and Remondino, F.},
  TITLE = {DEEP-IMAGE-MATCHING: A TOOLBOX FOR MULTIVIEW IMAGE MATCHING OF COMPLEX SCENARIOS},
  JOURNAL = {The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences},
  VOLUME = {XLVIII-2/W4-2024},
  YEAR = {2024},
  PAGES = {309--316},
  DOI = {10.5194/isprs-archives-XLVIII-2-W4-2024-309-2024}
}

@article{morelli2022photogrammetry,
  title={PHOTOGRAMMETRY NOW AND THEN--FROM HAND-CRAFTED TO DEEP-LEARNING TIE POINTS--},
  author={Morelli, Luca and Bellavia, Fabio and Menna, Fabio and Remondino, Fabio},
  journal={The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences},
  volume={48},
  pages={163--170},
  year={2022},
  publisher={Copernicus GmbH}
}

@article{ioli2024,
  title={Deep Learning Low-cost Photogrammetry for 4D Short-term Glacier
Dynamics Monitoring},
  author={Ioli, Francesco and Dematteis, Nicolò and Giordan, Daniele and Nex, Francesco and Pinto Livio},
  journal={PFG – Journal of Photogrammetry, Remote Sensing and Geoinformation Science},
  year={2024},
  DOI = {10.1007/s41064-023-00272-w}
}

Depending on the options used, consider citing the corresponding work of:

• KORNIA
• HLOC
• COLMAP
• OpenMVG
• MICMAC
• used local features and matchers