IAMAP: Unlocking Deep Learning in QGIS for non-coders and limited computing resources

Abstract

We introduce IAMAP, a user-friendly plugin for the geographic information system QGIS that allows users to (1) tile raster images and feed them into pre-trained deep learning model to extract image features; (2) reduce feature dimensionality; (3) perform clustering on features or their reduced representations; (4) generate feature similarity maps; (5) calibrate and validate supervised machine learning models to create maps. By enabling non-AI specialists to leverage the high-quality features provided by foundation models without requiring GPU capacity or extensive reference datasets, IAMAP contributes to the democratization of these computationally efficient and energy-conscious methods. Development of the plugin is opensource on GitHub. Documentation and tutorials are available on Read the Docs.

Machine learning assessment of industrial corrosion and public health risks in data limited aquifers of Northeastern Algeria
Mohamed Hamdy Eid, Omar Saeed, Péter Szűcs +10
Scientific Reports10 Oct 2026Peer-reviewedCC BY-NC-ND 4.0Full textHydrology & soilsCrossref
Physics-Informed Machine Learning in Hydrology and Water Resources: A Systematic Bibliometric Analysis
George Kwame Frimpong, Omer Mermer, Ibrahim Demir
EarthArXiv5 Oct 2026PreprintCC BY 4.0Full textHydrology & soilsEarthArXiv
Tropical Cyclone-Following Dataset for Limited-Area Machine Learning Applications
Minh-Khanh Luong, Chanh Kieu
arXiv2 Nov 2026PreprintCC BY 4.0Full textAtmospheric science & meteorologyarXiv
Hydrogen Digital Product Passport: Unlocking Trust and Transparency for a Global Low-Carbon Future.
David Parra, Lourdes F Vega, Magdy M Hassan +2
Environmental science & technology1 Sep 2026Peer-reviewedCC BY-NC-NDFull textEnvironmental science & pollutionEurope PMC
Unlocking Climate Adaptation at COP31
Enoch Ntsiful, François Cohen, Patrick Fahr
Les Cahiers Espace, Environnement, Risques et Résilience2 Nov 2026Peer-reviewedCC BY-SA 4.0Full textEnvironmental science & pollutionHAL

Cite

BibTeX RIS

BibTeX

@article{tresson2026iamap,
  title = {{IAMAP: Unlocking Deep Learning in QGIS for non-coders and limited computing resources}},
  author = {Paul Tresson and Pierre Le Coz and Hadrien Tulet and Anthony Malkassian and Maxime Réjou-Méchain},
  journal = {Journal of Open Source Software},
  year = {2026},
  month = jun,
  doi = {10.21105/joss.10329},
  eprint = {2508.00627},
  archivePrefix = {arXiv},
  url = {https://hal.science/hal-05793642v1},
  note = {CC BY 4.0},
}

RIS

TY  - JOUR
TI  - IAMAP: Unlocking Deep Learning in QGIS for non-coders and limited computing resources
AU  - Paul Tresson
AU  - Pierre Le Coz
AU  - Hadrien Tulet
AU  - Anthony Malkassian
AU  - Maxime Réjou-Méchain
PY  - 2026
DA  - 2026/06/25
JO  - Journal of Open Source Software
DO  - 10.21105/joss.10329
UR  - https://hal.science/hal-05793642v1
AB  - We introduce IAMAP, a user-friendly plugin for the geographic information system QGIS that allows users to (1) tile raster images and feed them into pre-trained deep learning model to extract image features; (2) reduce feature dimensionality; (3) perform clustering on features or their reduced representations; (4) generate feature similarity maps; (5) calibrate and validate supervised machine learning models to create maps. By enabling non-AI specialists to leverage the high-quality features provided by foundation models without requiring GPU capacity or extensive reference datasets, IAMAP contributes to the democratization of these computationally efficient and energy-conscious methods. Development of the plugin is opensource on GitHub. Documentation and tutorials are available on Read the Docs.
SN  - 1942-3926
ER  -

Abstract

Related

Cite