Laser remote sensing for land use inventory
By horoma AI Inc. | Published on February 17, 2022
Problem and objectives
Problem
- Every five years, Quebec’s regional county municipalities (“RCMs”) must produce an inventory showing the provincial government’s land use. The work requires a manual effort of approximately 500 hours.
Objectives
- Horoma AI’s goal was to develop, from February to December 2020, an automation tool for classifying land use for four Eastern Townships RCMs in southern Québec and the City of Sherbrooke.
Proposed solution
Solution
Horoma AI specializes in artificial intelligence (AI) and geomatics. Using remote sensing methods, the company automated the detailed land inventory process for the City of Sherbrooke and four RCMs in response to requests from the Ministère de l’Énergie et des Ressources naturelles du Québec (“MERN”). The mandate initially started as a research and development project.
Principle
Horoma AI first analyzed aerial optical images that provide information on the shape of the land. These multi-spectral images are collected by aircraft using optical laser remote sensing. Second, the company developed algorithms that analyze the data in the resulting images of the land and classify the land according to the categories required for the report to be submitted to the Government of Québec. For this purpose, the land was divided into about 20 classes, including forests, wetlands, buildings and several others.
Role of AI
Horoma AI used the following techniques for the project:
- Machine learning;
- Stacking;
- Deep learning;
- Deep neural networks;
- Convolutional neural networks;
- Classification;
- Clustering.
Impacts
Main outcomes
The technology developed by Horoma AI is used to automate close to 90% of the work required to prepare the report for MERN. Using the platform that was developed saves the participating municipalities significant time. According to Yvan Ouellet, President and CEO of Horoma AI, their technology enables specific results to be attained and fieldwork to be optimized, as well as allow geomatics specialists, foresters or anyone in related fields to optimize fieldwork. In addition to gains in efficiency, automation increases computation and image analysis capacity. The model is capable of ensuring the integrity and repeatability of results. Experts can then use the results of analyses developed using AI as support in decision-making.
Impact on RCMs and the City of Sherbrooke
- Reduction in time spent preparing the MRN report from 500 to 72 hours;
- Increased accuracy of collected data;
- Potential for more in-depth data analysis.
Impact on Horoma AI
- Experience in land classification;
- Increase in technological maturity.
Generation and dissemination of new knowledge
- Training provided by Horoma AI to geomatics specialists from the RCMs that participated in the project.
Challenges takled
- Availability of sufficient data to train a model;
- Hiring personnel with expertise in geomatics and AI.
Conditions for success
Team mobilised
Internal resources
- Data engineers;
- Software developers;
- Geomatics specialists.
External resources
- Finance manager;
- Geomatics team project managers;
- Université de Sherbrooke’s Centre d’applications et de recherche en télédétection (CARTEL);
- Computer Research Institute of Montréal (CRIM).
Collaborations
During its various projects, Horoma AI developed partnerships with organizations at the local, national and international levels. These professional relationships include:
- Institut de recherche d’Hydro-Québec (IREQ);
- NorthStar Earth & Space Inc.; LBM Agtech;
- Grupo Energisa (Brazil).
Funding received
Grant received from the National Research Council of Canada’s Industrial Research Assistance Program (NRC-IRAP).
Project stages
Project definition (discussions with client to determine the potential objectives and issues);
Development of the model (ongoing interactions with the client);
Project conclusion (lessons learned).
Project stages - to come
The company plans on continuing to develop the project in view of integrating a SaaS platform and solution.
Context of the solution
Customers
Horoma AI’s services are intended for businesses and governments in the following fields:
- Farming, forestry, fishing and hunting;
- Utilities;
- Professional, scientific and technical services.