Detection of invasive plant species

Client

Quebec City

Objective

Through its Natural Environment Integrity Program, Quebec City ensures that the natural environments it owns retain their natural character. To achieve this, a number of actions are planned, including the fight against invasive plant species, for which the City wishes to adopt a ten-year management plan (2024-2033). As such, it needs to carry out an inventory of invasive plant species on its territory.

The City therefore wishes to evaluate the possibilities of developing a methodology using remote sensing from drone images to map/inventory invasive plant species.

Methodology

Our work consists in developing deep learning models for the identification and mapping of invasive plant species from drone images. These initial models will be trained and validated with images acquired in autumn 2023 by the CGQ team.

The models developed will aim to identify each of the four invasive plant species groups targeted in this study: giant hogweed, knotweed (Japanese, Bohemian, Sakhalin), buckthorn (borage and cathartic) and common reed.

Equipment

Matrice 600 Pro drone

Phase One camera

Multispectral camera

Field

Environmental monitoring