Adapting a global plant identification model to detect invasive alien plant species in high-resolution road side images

Early detection of invasive alien plant species is crucial for addressing their environmental impact. Recent advancements in vehicle-mounted equipment enable automatic analysis of high-resolution images to detect invasive plants along roadsides, a primary vector for their spread. Deep learning techn...

Full description

Saved in:
Bibliographic Details
Main Authors: Vincent Espitalier, Jean-Christophe Lombardo, Hervé Goëau, Christophe Botella, Toke Thomas Høye, Mads Dyrmann, Pierre Bonnet, Alexis Joly
Format: Article
Language:English
Published: Elsevier 2025-11-01
Series:Ecological Informatics
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S1574954125001384
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850123759627796480
author Vincent Espitalier
Jean-Christophe Lombardo
Hervé Goëau
Christophe Botella
Toke Thomas Høye
Mads Dyrmann
Pierre Bonnet
Alexis Joly
author_facet Vincent Espitalier
Jean-Christophe Lombardo
Hervé Goëau
Christophe Botella
Toke Thomas Høye
Mads Dyrmann
Pierre Bonnet
Alexis Joly
author_sort Vincent Espitalier
collection DOAJ
description Early detection of invasive alien plant species is crucial for addressing their environmental impact. Recent advancements in vehicle-mounted equipment enable automatic analysis of high-resolution images to detect invasive plants along roadsides, a primary vector for their spread. Deep learning technologies show promise for processing this data efficiently, but the choice of approach significantly affects both computational and human resource costs. Object detection and segmentation methods require costly annotations, making them impractical for scaling to the thousands of invasive species worldwide. In contrast, multi-label classification, i.e. to predict all species present in the image, is less demanding but still challenging to implement without many annotated images for numerous species. However, large datasets from citizen science platforms such as Pl@ntNet or iNaturalist offer rich visual data for classifying individual plant species. In this article, we assess whether large plant identification models trained on such data can be leveraged for species detection in high-resolution images. Specifically, we explore two approaches: a multi-label classification model and a tiling-based model, using a vision transformer from the Pl@ntNet platform. We evaluate these models on high-resolution roadside images, both using a pre-trained model without fine-tuning and after applying fine-tuning. Our findings indicate that the tiling approach significantly outperforms other methods without fine-tuning and shows a slight advantage when fine-tuning is applied, demonstrating significant potential for detecting thousands of species without task-specific adaptation.
format Article
id doaj-art-62685644afc0428290b8e9dc4fbf6c39
institution OA Journals
issn 1574-9541
language English
publishDate 2025-11-01
publisher Elsevier
record_format Article
series Ecological Informatics
spelling doaj-art-62685644afc0428290b8e9dc4fbf6c392025-08-20T02:34:31ZengElsevierEcological Informatics1574-95412025-11-018910312910.1016/j.ecoinf.2025.103129Adapting a global plant identification model to detect invasive alien plant species in high-resolution road side imagesVincent Espitalier0Jean-Christophe Lombardo1Hervé Goëau2Christophe Botella3Toke Thomas Høye4Mads Dyrmann5Pierre Bonnet6Alexis Joly7CIRAD, UMR AMAP, Montpellier, Occitanie, FranceInria, LIRMM, Univ Montpellier, CNRS, Montpellier, FranceCIRAD, UMR AMAP, Montpellier, Occitanie, France; Corresponding author.Inria, LIRMM, Univ Montpellier, CNRS, Montpellier, FranceDepartment of Ecoscience and Arctic Research Centre, Aarhus University, C. F. Møllers Allé 8, DK 8000, Aarhus C, DenmarkAI Lab ApS, Aarhus, 8210, DenmarkCIRAD, UMR AMAP, Montpellier, Occitanie, FranceInria, LIRMM, Univ Montpellier, CNRS, Montpellier, FranceEarly detection of invasive alien plant species is crucial for addressing their environmental impact. Recent advancements in vehicle-mounted equipment enable automatic analysis of high-resolution images to detect invasive plants along roadsides, a primary vector for their spread. Deep learning technologies show promise for processing this data efficiently, but the choice of approach significantly affects both computational and human resource costs. Object detection and segmentation methods require costly annotations, making them impractical for scaling to the thousands of invasive species worldwide. In contrast, multi-label classification, i.e. to predict all species present in the image, is less demanding but still challenging to implement without many annotated images for numerous species. However, large datasets from citizen science platforms such as Pl@ntNet or iNaturalist offer rich visual data for classifying individual plant species. In this article, we assess whether large plant identification models trained on such data can be leveraged for species detection in high-resolution images. Specifically, we explore two approaches: a multi-label classification model and a tiling-based model, using a vision transformer from the Pl@ntNet platform. We evaluate these models on high-resolution roadside images, both using a pre-trained model without fine-tuning and after applying fine-tuning. Our findings indicate that the tiling approach significantly outperforms other methods without fine-tuning and shows a slight advantage when fine-tuning is applied, demonstrating significant potential for detecting thousands of species without task-specific adaptation.http://www.sciencedirect.com/science/article/pii/S1574954125001384Invasive speciesRoadside biodiversity monitoringPlant recognitionDeep learningLarge-scale species identificationTiling
spellingShingle Vincent Espitalier
Jean-Christophe Lombardo
Hervé Goëau
Christophe Botella
Toke Thomas Høye
Mads Dyrmann
Pierre Bonnet
Alexis Joly
Adapting a global plant identification model to detect invasive alien plant species in high-resolution road side images
Ecological Informatics
Invasive species
Roadside biodiversity monitoring
Plant recognition
Deep learning
Large-scale species identification
Tiling
title Adapting a global plant identification model to detect invasive alien plant species in high-resolution road side images
title_full Adapting a global plant identification model to detect invasive alien plant species in high-resolution road side images
title_fullStr Adapting a global plant identification model to detect invasive alien plant species in high-resolution road side images
title_full_unstemmed Adapting a global plant identification model to detect invasive alien plant species in high-resolution road side images
title_short Adapting a global plant identification model to detect invasive alien plant species in high-resolution road side images
title_sort adapting a global plant identification model to detect invasive alien plant species in high resolution road side images
topic Invasive species
Roadside biodiversity monitoring
Plant recognition
Deep learning
Large-scale species identification
Tiling
url http://www.sciencedirect.com/science/article/pii/S1574954125001384
work_keys_str_mv AT vincentespitalier adaptingaglobalplantidentificationmodeltodetectinvasivealienplantspeciesinhighresolutionroadsideimages
AT jeanchristophelombardo adaptingaglobalplantidentificationmodeltodetectinvasivealienplantspeciesinhighresolutionroadsideimages
AT hervegoeau adaptingaglobalplantidentificationmodeltodetectinvasivealienplantspeciesinhighresolutionroadsideimages
AT christophebotella adaptingaglobalplantidentificationmodeltodetectinvasivealienplantspeciesinhighresolutionroadsideimages
AT tokethomashøye adaptingaglobalplantidentificationmodeltodetectinvasivealienplantspeciesinhighresolutionroadsideimages
AT madsdyrmann adaptingaglobalplantidentificationmodeltodetectinvasivealienplantspeciesinhighresolutionroadsideimages
AT pierrebonnet adaptingaglobalplantidentificationmodeltodetectinvasivealienplantspeciesinhighresolutionroadsideimages
AT alexisjoly adaptingaglobalplantidentificationmodeltodetectinvasivealienplantspeciesinhighresolutionroadsideimages