GeoImageNet: artificial intelligence applied to satellite images!

GeoImageNet: artificial intelligence applied to satellite images!

Published April 29, 2019
For immediate release



GeoImageNet: artificial intelligence applied to satellite images!

Montréal, April 29, 2019 – CRIM is proud to be part of an innovative project that will support the work of Canadian geospatial researchers to accelerate innovation in this leading-edge sector.

The GeoImageNet project, led by Yacine Bouroubi and Samuel Foucher, respectively professor of applied geomatics at Université de Sherbrooke and senior researcher in remote sensing at CRIM, aims to create a new research platform for the collaborative annotation of satellite images as a tool for researchers and organizations in the geospatial sector. This project is being carried out in collaboration with Effigis Géo-Solutions and other partners, such as Natural Resources Canada's Canada Centre for Mapping and Earth Observation (CCMEO). CANARIE supports this project through its Research Software Program.

CRIM will contribute its expertise in deep learning applied to satellite images and in the development of geospatial platforms. Effigis and the CCMEO will provide more than 10,000 km2 of very high resolution images to the team.

GeoImageNet will include tools for annotating very high spatial resolution (VHR) satellite images. The project aims to develop deep learning algorithms for land-use mapping and object detection from VHR satellite images.

GeoImageNet in a nutshell

  • GeoImageNet is a new research platform that incorporates some of the software components of the PAVICS platform, created by CRIM and OURANOS and specializing in the visualization of climate data. The core of the platform is based on OGC (Open Geospatial Consortium) standards.
  • The GeoImageNet project makes it possible to annotate very high resolution satellite images. For example, some tools could detect specific objects in the images or accurately map out the type of land present in a given area. 
  • GeoImageNet will allow the creation of a training database for deep learning and testing of models developed by experts in the field. The project will build a database of annotated images with a taxonomy adapted to the content of THR images. The training database will be composed of regions of interest including annotations and will be available for download by researchers as a deep learning database, while a portion will be kept for testing models once they have been uploaded to the platform. 
  • Once completed, a second GeoImageNet platform will be deployed at Université de Sherbrooke and will be open to the various Canadian organizations (university labs, research centres, industry, etc.) working in remote sensing and the application of artificial intelligence in this field, so that they can use it and contribute to it.

CRIM, a trusted ally for geospatial projects!

Since the launch of the PAVICS platform, CRIM's world-class reputation in the field of geospatial platform development is well established. GeoImageNet marks a new milestone that promises to optimize the work of researchers in the field, at a time when artificial intelligence offers the opportunity to exploit satellite data to its full potential.

Starting this summer, Canadian researchers in the field will be called upon to improve and test the first version of the platform. Stay tuned for updates on the call for participation!

About CRIM

CRIM is an applied research and expertise centre in information technology, dedicated to making organizations more effective and competitive through the development of innovative technology and the transfer of leading edge know-how, while contributing to scientific advancement.

It helps organizations, primarily SMBs, demystify and gain access to leading-edge technology, such as artificial intelligence, to efficiently address the technological challenges they face. Its IT researchers and professionals develop a wide array of applications in diverse areas and work in such fields of expertise as machine learning, computer vision, speech recognition, automatic natural language processing, data science and operational research.

CRIM is a non-profit organization whose neutrality and strong network make it an indispensable resource. Its work is in line with the policies and strategies of its major financial partner, the ministère de l'Économie et de l'Innovation.

– 30 –

Information :

Antonia Leney-Granger, communications agent
CRIM – Computer Research Institute of Montréal | | 514 840-1234

Upcoming event

  • Gala des Prix Innovation 2020 de l'ADRIQ
    19 November 2020 0:00
    Palais des Congrès de Montréal
    Le Gala Prix Innovation 2020 de l'ADRIQ aura lieu le 19 novembre 2020, au Palais des congrès de Montréal.
  • AIxSPACE RT @AIxSPACE_ca: Registrations are open for AIxSPACE! We're looking forward to meeting you all at the 1st event dedicated to #AI applied to…
  • Vous souhaitez rencontrer des grandes compagnies, chercheurs et startups afin d'avoir une discussion de fond sur la…