The presentation will showcase recent work in the agriculture and forestry field on leveraging machine learning in remote sensing to automate the detection and extraction of micro-level information over satellite imagery. Emphasis will also be placed on how the work was made possible by available open municipal and provincial data as well as how this information led to the dissemination of further open micro-level spatial data, specifically on greenhouses as part of the Statistics Canada's Linkable Open Data Environment (LODE).
The session will conclude with further information on the LODE as well as an open discussion on what areas users would like to see expanded (e.g., What else could be detected over satellite imagery and made available as open micro data? What other municipal information could be released as open data and be added to the LODE?)
Date: Tuesday, May 14, 2024, 1:30 to 2:30 PM EDT
About the presenters:
Jean Le Moullec, Manager, Image Analytics and Applied AI, Statistics Canada
Jean has been working in the area of official statistics, unstructured data and modelling for more than 12 years. Past domains of experience include the Consumer Price Index and measures of spatial accessibility as well as the exploration and integration of scanner data, web scraping, crowdsourcing and open data. Since joining the Data Science and Innovation Division in 2020, Jean’s work has focused on leading a team in executing projects related to:
- using machine learning for remote sensing of information from satellite imagery
- using high frequency and early survey data to nowcast
- economic statistics
- developing tools to more efficiently navigate unstructured HR data (e.g., Curriculum Vitaes and applications).
Throughout all these projects, the themes of innovation, project management, non-traditional data sources and analytics has been constant.
Sara Burns, Remote Sensing Analyst, AgZero Project, Agriculture Division, Statistics Canada
Sara has been working in the field remote sensing and GIS for 7 years following her bachelors and advanced degree.
During her time at Statistics Canada, she has worked inclusively within the Agriculture division, analysing geographic data sets for various agricultural applications for the department, including digitization, classification and emergency maps for agricultural operators. Saras focus has been leading projects with multi- disciplinary teams, including Data Scientists, to experiment with automatic detection and classification of various agricultural components on our surveys and census. This is done in the effort to forward our initiative in reducing survey response burden, eliminate manual work and provide timely data to our users through Statistics Canada AgZero Initiative:
- Preparing and analysing GIS and Earth observation data for machine learning models
- Use of earth observation and GIS data for various agricultural use cases
- Frequently experiments with various geographics data sets with new open-source programming languages and Google Earth Engine for analysis and ideas relating to the department's innovation calendar.
Marina Smailes, Unit Head, Data Exploration and Integration Lab, Centre for Special Business Projects, Statistics Canada
Marina has been working on open data and geospatial analysis at Statistics Canada for over 6 years and is one of the key contributors to the StatCan Linkable Open Data Environment. Other than running the development open databases for the Linkable Open Data Environment, her work focuses on using open and alternative data inputs for experimental indicator development for clients across the federal government. Some previous projects involved:
- Sustainable Development Goal indicators
- Environmental equity research
- Proximity measures to local services
Marina and the Data Exploration and Integration Lab aim to highlight open data and open-source tools for disaggregated data development.
Number of attendees: 37
Downloads:
Recording: