A5085793980- Species Distribution and Climate Change
- Smart Agriculture and AI
- Plant and animal studies
- French Literature and Criticism
- Genomics and Phylogenetic Studies
- Renaissance Literature and Culture
- Remote Sensing in Agriculture
- Semantic Web and Ontologies
- Carcinogens and Genotoxicity Assessment
- Advanced Image and Video Retrieval Techniques
- Geographic Information Systems Studies
- Research Data Management Practices
- Plant Pathogens and Fungal Diseases
- Plant Pathogens and Resistance
- Identification and Quantification in Food
- Ecology and Vegetation Dynamics Studies
- Animal and Plant Science Education
- Image Retrieval and Classification Techniques
- Heme Oxygenase-1 and Carbon Monoxide
- Effects and risks of endocrine disrupting chemicals
- Agriculture and Rural Development Research
- Wildlife Ecology and Conservation
- Animal Vocal Communication and Behavior
- Toxic Organic Pollutants Impact
- Plant and Fungal Species Descriptions
Centre de Coopération Internationale en Recherche Agronomique pour le Développement
2016-2025
UMR Botanique et Modélisation de l’Architecture des Plantes et des végétations
2015-2024
Agropolis International
2014-2024
Université de Montpellier
2018-2024
Institut de Recherche pour le Développement
2018-2024
Centre National de la Recherche Scientifique
2005-2024
Institut National de Recherche pour l'Agriculture, l'Alimentation et l'Environnement
1998-2024
Institut Agro Montpellier
2018-2024
Université Grenoble Alpes
2018-2024
Laboratoire Bordelais de Recherche en Informatique
2024
The control of plant diseases is a major challenge to ensure global food security and sustainable agriculture. Several recent studies have proposed improve existing procedures for early detection through modern automatic image recognition systems based on deep learning. In this article, we study these methods in detail, especially those convolutional neural networks. We first examine whether it more relevant fine-tune pre-trained model identification task rather than general object task....
Hundreds of herbarium collections have accumulated a valuable heritage and knowledge plants over several centuries. Recent initiatives started ambitious preservation plans to digitize this information make it available botanists the general public through web portals. However, thousands sheets are still unidentified at species level while numerous should be reviewed updated following more recent taxonomic knowledge. These annotations revisions require an unrealistic amount work for carry out...
Speeding up the collection and integration of raw botanical observation data is a crucial step towards sustainable development agriculture conservation biodiversity. Initiated in context citizen sciences project, main contribution this paper an innovative collaborative workflow focused on image-based plant identification as mean to enlist new contributors facilitate access data. Since 2010, hundreds thousands geo-tagged dated photographs were collected revised by novice, amateur expert...
Premise Weed removal in agriculture is typically achieved using herbicides. The use of autonomous robots to reduce weeds a promising alternative solution, although their implementation requires the precise detection and identification crops allow an efficient action. Methods We trained evaluated instance segmentation convolutional neural network aimed at segmenting identifying each plant specimen visible images produced by agricultural robots. resulting data set comprised field on which...
Machine learning (ML) has great potential to drive scientific discovery by harvesting data from images of herbarium specimens-preserved plant material curated in natural history collections-but ML techniques have only recently been applied this rich resource. particularly strong prospects for the study phenological events such as growth and reproduction. As a major indicator climate change, driver ecological processes, critical determinant fitness, phenology is an important frontier...
Convolutional Neural Networks (CNNs) are statistical models suited for learning complex visual patterns. In the context of Species Distribution Models (SDM) and in line with predictions landscape ecology island biogeography, CNN could grasp how local structure affects prediction species occurrence SDMs. The can thus reflect signatures entangled ecological processes. Although previous machine-learning based SDMs learn influences environmental predictors, they cannot acknowledge influence...
This paper presents a synthesis of ImageCLEF 2013 plant identification task, system-oriented testbed dedicated to the evaluation image-based technologies. With 12 participating groups coming from over 9 countries and 33 submitted runs, campaign confirmed increasing interest multimedia community in ecology-related challenges (respectively 10 11 crossed finish line 2011 2012). Contrary two previous years that were exclusively focused on leaf images, coverage task was extended six different...
Plant diseases have a significant impact on global food security and the world's agricultural economy. Their early detection classification increase chances of setting up effective control measures, which is why search for automatic systems that allow this major interest to our society. Several recent studies reported promising results in plant from RGB images basis Convolutional Neural Networks (CNN). These been successfully experimented large number crops symptoms, they shown advantages...
Premise of the Study Phenological annotation models computed on large‐scale herbarium data sets were developed and tested in this study. Methods Herbarium specimens represent a significant resource with which to study plant phenology. Nevertheless, phenological is time‐consuming, requires substantial human investment, difficult mobilize at large taxonomic scales. We created evaluated new methods based deep learning techniques automate stages these four representing temperate, tropical,...
Abstract The hypersensitive response and systemic acquired resistance (SAR) can be induced in tobacco (Nicotiana tabacum L.) plants by cryptogein, an elicitin secreted Phytophthora cryptogea. Stem application of cryptogein leads to the establishment subsequent leaf infection with parasitica var nicotianae, agent black shank disease. We have studied early events that occur after show here a gene encoding extracellular S-like RNase NE is expressed inoculation pathogenic fungus. Upon induction...
[email protected] is an image sharing and retrieval application for the identification of plants, available on iPhone iPad devices. Contrary to previous content-based applications it can work with several parts plant including flowers, leaves, fruits bark. It also allows integrating user's observations in database thanks a collaborative workflow involving members social network specialized plants. Data collected so far makes one largest mobile tool.
Premise of the Study A species distribution model computed with automatically identified plant observations was developed and evaluated to contribute future ecological studies. Methods We used deep learning techniques identify opportunistic made by citizens through a popular mobile application. compared modeling invasive alien plants based on these data inventories experts. Results The trained models have reasonable predictive effectiveness for some species, but they are biased massive...
Abstract 1. Successful monitoring and management of plant resources worldwide needs the involvement civil society to support natural reserve managers. Because it is difficult correctly quickly identify species for non‐specialists, development recent techniques based on automatic visual identification should facilitate increase public engagement in citizen science initiatives. 2. Automatic platforms are new most scientists land Pl@ntNet such a platform, available since 2013 web mobile...
The use of naturalist mobile applications have dramatically increased during last years, and provide huge amounts accurately geolocated species presences records. Integrating this novel type data in distribution models (SDMs) raises specific methodological questions. Presence-only SDM methods require background points, which should be consistent with sampling effort across the environmental space to avoid bias. A standard approach is uniformly distributed points (UB). When multiple are...