Francis Charette Migneault

Francis Charette Migneault
Centre de recherche informatique de Montréal | CRIM · Vision et imagerie

Master's Degree in Automation Engineering

About

20
Publications
17,445
Reads
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
2
Citations
Introduction
I work on artificial vision and machine learning algorithms over multiple fields of study including geospatial satellite imagery, object detection and tracking, and face recognition frameworks employed in security and camera surveillance systems. I employ various methods, classifiers, libraries and development technologies to realize my work. Most of my projects consist in testing the robustness of existing methodologies in the literature, implementing some of them, potentially improving algorithms using variations of classification and image processing procedures, as well as packaging ML applications for production server environments.
Additional affiliations
March 2018 - present
Centre de recherche informatique de Montréal
Position
  • Developer
April 2015 - August 2015
École de Technologie Supérieure
Position
  • Artificial vision laboratory evaluator (GPA659)
Description
  • - Evaluation and correction of laboratory reports from students. - Demonstration of artificial vision procedures and concepts for laboratory manipulations.
Education
January 2016 - December 2018
École de Technologie Supérieure
Field of study
  • Automated Production Engineering, concentration in Intelligent Systems
September 2011 - August 2015
École de Technologie Supérieure
Field of study
  • Automated Production Engineering

Publications

Publications (20)
Presentation
We present here the recent progress of the DACCS (Data Analytics for Canadian Climate Services) project [1], funded by the Canadian Foundation for Innovation (FCI) and various provincial partners in Canada, in response to a national cyberinfrastructure challenge. The project’s development phase is set to finish in 2023, while funding for maintenanc...
Article
https://medium.com/crim/contributing-to-libtorch-recent-architectures-and-vanilla-training-pipeline-3789c7bf6959 In August 2021, a PR aimed at adding a SOTA architecture (namely EfficientNet) to TorchVision, a Python-based PyTorch package for computer vision experiments, was submitted on GitHub. Even though deep learning practitioners are used to...
Technical Report
Full-text available
Platforms for the exploitation of Earth Observation (EO) data have been developed by public and private companies in order to foster the usage of EO data and expand the market of Earth Observation-derived information. A fundamental principle of the platform operations concept is to move the EO data processing service’s user to the data and tools, a...
Conference Paper
Data privacy issues and regulations are on the rise. Privacy and transparency concerns are quickly defining a new era in data management and service operation across applications, platforms and industries. With a continuous flow of revisited privacy laws worldwide, adopted in different regulations such as the General Data Protection Regulation (GDP...
Technical Report
Full-text available
The CCMEO is currently working on implementing an inference service to extract mapping elements using imaging and support several DL models. One of the components of that service is researching a trained model. To be able to use a model on new images, we need to be able to determine the context in which a model was trained (type of images and their...
Technical Report
Full-text available
For several years now, CRIM has been developing research software for its own researchers, and for the larger Canadian scientific community. A key milestone in this series of initiatives is project PAVICS (Plateforme pour l’Analyse et la Visualisation de l’Information Climatique et Scientifique), funded by CANARIE, and aiming to facilitate the Big...
Technical Report
Full-text available
This Open Geospatial Consortium (OGC) Engineering Report (ER) describes the advancement of an Execution Management System (EMS) to support Web Processing Service (WPS) climate processes deployed on the Earth System Grid Federation (ESGF). The report introduces climate data, processes and applications into Common Workflow Language (CWL) workflows wi...
Conference Paper
Full-text available
Earth Observations (EO) enable scientific research, such as the study of meteorology and climate, ecosystems and forests, hydrology and marine life. Applications of EO help protect populations from disasters and improve life in intelligent cities. Increasingly, Machine Learning techniques are seen as key to solve these complex multidisciplinary pro...
Poster
Full-text available
A collaborative initiative between remote sensing researchers, web platform developers, artificial intelligence experts and satellite imagery providers. Implement a web platform that allows users to make annotations on VHR images (e.g.: Pleiades 50 cm; WorldView-2 40 cm; WorldView-3 30 cm). Create large datasets of annotations and related patches....
Conference Paper
Full-text available
Still-to-video face recognition (FR) is an important function in many video surveillance applications, allowing to recognize target individuals of interest appearing over a distributed network of cameras. Systems for still-to-video FR match faces captured in videos under challenging conditions against facial models, often based on a single referenc...
Thesis
Full-text available
La reconnaissance de visages (FR, Face Recognition) est une fonction importante des systèmes de vidéosurveillance (FRiVS, FR in Video Surveillance) pour permettre la vérification et l’identification d’individus d’intérêt qui apparaissent dans une scène capturée par un réseau distribué de caméras. Une application recherchée des systèmes de FRiVS est...
Poster
Full-text available
Evaluation of road detection from synthetic aperture radar (SAR) satellite images for eventually obtaining a map of the road network conditions to better guide emergency response teams for disaster interventions.
Conference Paper
Full-text available
Étant donné que les systèmes de surveillance deviennent de plus en plus étendus dans les milieux urbanisés, l’apport de techniques automatisées performantes est requis. Cette automatisation permet de répondre à l’accroissement du nombre de caméras présentes dans les systèmes tout en accommodant la quantité croissante d’individus impliqués dans l’en...
Poster
Full-text available
Comparative evaluation of three classifiers (Nearest Neighbor, Fuzzy-ARTMAP and Support Vector Machines) for face recognition with single sample per person employed for training. The study employs Viola-Jones face detection and local binary pattern (LBP) descriptors to train each classifier. Four different tests are executed to evaluate the recogni...
Poster
Full-text available
Evaluation of image preprocessing operations to allow automatic road detection from radar satellite images. The project focuses on evaluating best filters to reduce radar speckle noise, extract road segments from the image and recombining them to detect a complete road network. This projet aimed at identifying damaged roads to establish a suitable...

Questions

Question (1)
Hi,
I have a colleague which is planning to work on such kind of problematic and data evaluation for work absenteeism. I would like to know if you have progressed on this projet? Is there any interesting dataset or algorithm that you have already looked into that seems to have more potential?
Thank you

Network

Cited By

Projects

Projects (8)
Project
PAVICS is a web platform that provides access to climate data and services. Users can launch analytical processes on remote servers, out-sourcing the technical, tedious and/or compute intensive work.
Project
The Data Analytics for Canadian Climate Services (DACCS) gateway will facilitate the conversion of raw climate data from satellite observations and simulation models into relevant, credible and actionable information products. DACCS is a Science Gateway comprised of reusable modular components that will provide services to a broad scientific community through a unified user interface for: dataset search and discovery; user data management; data analytics; visualisation; workflow management; publishing; and application deployment. As part of a larger Federated Cloud Infrastructure, through this gateway, scientists can access data from different sources and process them using existing algorithms developed by the academic, government and private sectors. Typical outputs are tables, time series and graphics that support climate change impact studies, vulnerability assessments, and decision-making.
Archived project
What : GeoImageNet is a unique collaborative initiative involving remote sensing researchers, developers of digital research platforms, artificial intelligence experts and professionals dedicated to derive value from satellite imagery. How : By facilitating the creation and download of annotations on Pleiades (50 cm) images. The imagery used to build this database includes more than 10,000 km2 of Pleiades images covering Canada's major cities as well as various other natural and anthropogenic environments (forests, wetlands, mining sites, agricultural areas, etc.). These annotations are based on a taxonomy containing many objects (approx. 180) and land cover types (approx. 50). Why : To promote deep learning research on Earth Observation (EO) data for detection, segmentation and other automated tasks. This will allow researchers from diverse institutions to collaborate in a more structured and effective manner for the application of deep learning in remote sensing and to develop new value-added products based on VHR satellite images. This synergy will facilitate making more progress in research, both in remote sensing applications and in the development of machine learning algorithms.