Pl@ntNet Crops: merging citizen science observations and structured survey data to improve crop recognition for agri-food-environment applications

Velde, Marijn van der; Goëau, Hervé; Bonnet, Pierre; d’Andrimont, Raphaël; Yordanov, Momtchil; Affouard, Antoine; Claverie, Martin; Czúcz, Bálint; Elvekjaer, Neija; Martinez-Sanchez, Laura; Rotllàn‐Puig, Xavier; Sima, Aleksandra; Verhegghen, Astrid; Joly, Alexis

doi:10.1088/1748-9326/acadf3

Cited by 6 publications

(4 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Given the continuous growth in data gathered via crowdsourcing, an unprecedented amount of geo-tagged and time-stamped photographs information is now collected using free apps for smartphones. [19][20][21] These data can aid in enhancing the comparison between ground-based and satellitebased LC information.…”

Section: Problem Formulationmentioning

confidence: 99%

Accuracy assessment of land-use-land-cover maps: the semantic gap between in situ and satellite data

Paris,

Martinez-Sanchez,

van der Velde

et al. 2023

Image and Signal Processing for Remote Sensing XXIX

View full text Add to dashboard Cite

The availability of high-resolution, open, and free satellite data has facilitated the production of global Land-Use-Land-Cover (LULC) maps, which are extremely important to monitor the Earth’s surface constantly. However, generating these maps demands significant efforts in collecting a vast amount of data to train the classifier and to assess their accuracy. Although in-situ surveys are generally regarded as reliable sources of information, it is important to note that there may be inconsistencies between the in-situ data and the information derived from satellite data. This can be attributed to various factors (1) differences in viewpoint perspectives, i.e., aerial versus ground views, and (2) spatial resolution of the satellite images versus the extent of the Land-Cover (LC) present in the scene. The aim of this paper is to explore the feasibility of using geo-referenced street-level imagery to bridge the gap between information provided by field surveys and satellite data. Unlike conventional in-situ surveys that typically provide geo-tagged location-specific information on LULC, street-level images offer a richer semantic context for the sampling point under examination. This allows for (1) an improved interpretation of LC characteristics, and (2) a stronger correlation with satellite data. The experimental analysis was conducted considering the 2018 Land Use and Coverage Area Frame Survey (LUCAS) in-situ data, the LUCAS landscape (street-level) images and three high-resolution thematic products derived from satellite data, namely, Google’s Dynamic World, ESA’s World Cover, and Esri’s Land Cover maps.

show abstract

Section: Problem Formulationmentioning

confidence: 99%

Accuracy assessment of land-use-land-cover maps: the semantic gap between in situ and satellite data

Paris,

Martinez-Sanchez,

van der Velde

et al. 2023

Image and Signal Processing for Remote Sensing XXIX

View full text Add to dashboard Cite

show abstract

“…The diversity of human behavior and electronic devices makes citizen science-based plant photographs very heterogeneous. This can be a challenge for deep learning applications, such as species recognition or plant trait characterization (Schiller et al, 2021;Van Horn et al, 2021;van Der Velde et al, 2023;Affouard et al, 2017), where models have to identify features that hold across various viewing angles, distances, or illumination conditions. However, this heterogeneity might also be of great value, given that citizens depict the appearance of plants under various site, environmental, and phenological conditions.…”

Section: Model Transferability Across Seasons and Orthoimage Acquisit...mentioning

confidence: 99%

From simple labels to semantic image segmentation: Leveraging citizen science plant photographs for tree species mapping in drone imagery

Soltani,

Ferlian,

Eisenhauer

et al. 2023

Preprint

View full text Add to dashboard Cite

Abstract. Knowledge of plant species distributions is essential for various applications, such as nature conservation, agriculture, and forestry. Remote sensing data, especially high-resolution orthoimages from Unoccupied Aerial Vehicles (UAVs), were demonstrated to be an effective data source for plant species mapping. Particularly, in concert with novel pattern recognition methods, such as Convolutional Neural Networks (CNNs), plant species can be accurately segmented in such high-resolution UAV images. Training such pattern recognition models for species segmentation that are transferable across various landscapes and remote sensing data characteristics often requires excessive training data. Training data are usually derived in the form of segmentation masks from field surveys or visual interpretation of the target species in remote sensing images. Still, both methods are laborious and constrain the training of transferable pattern recognition models. Alternatively, pattern recognition models could be trained on the open knowledge of how plants look as available from smartphone-based species identification apps, that is, millions of citizen science-based smartphone photographs and the corresponding species label. However, these pairs of citizen science-based photographs and simple species labels (one label for the entire image) cannot be used directly for training state-of-the-art segmentation models used for UAV image analysis, which require per-pixel labels for training (also called masks). Here, we overcome the limitation of simple labels of citizen science plant observations with a two-step approach: In the first step, we train CNN-based image classification models using the simple labels and apply them in a moving-window approach over UAV orthoimagery to create segmentation masks. In the second phase, these segmentation masks are used to train state-of-the-art CNN-based image segmentation models with an encoder-decoder structure. We tested the approach on UAV orthoimages acquired in summer and autumn on a test site comprising ten temperate deciduous tree species in varying mixtures. Several tree species could be mapped with surprising accuracy (mean F1-score = 0.47). In homogenous species assemblages, the accuracy increased considerably (mean F1-score 0.55). The results indicate that many tree species can be mapped without generating training data and by integrating pre-existing knowledge from citizen science. Moreover, our analysis revealed that citizen science photographs’ variability in acquisition data and context facilitates the generation of models that are transferable through the vegetation season. Thus, citizen science data may greatly advance our capacity to monitor hundreds of plant species and, thus, Earth's biodiversity across space and time.

show abstract

“…Land cover classification examples: artificial land, built areas or covered buildings separated into three categories ≤3 floors or less than 10 m high, > 3 floors or higher than 10 m, greenhouses), and unbuilt areas (with artificial cover) [27][28][29]59].…”

Section: Analysis Of Land Coveragementioning

confidence: 99%

“…Cultivated land means there is plant production on a land plot (cultivated land, the crop has not yet appeared in a plowed and sown field, crop already harvested that is identifiable by plant residues, crop already harvested that is not identifiable by plant residues, species inclusion of plants, mixed crops, a mixture of two cereals, mixtures of more than two cereals, uncovered areas between the rows of an orchard, grassy areas between the rows of a vineyard, fruit trees in vegetable gardens, isolated fruit trees, chestnuts, cherries, walnuts and other fruit-bearing trees, and abandoned crop areas) [27][28][29]54,59,60].…”

Section: Analysis Of Land Usementioning

confidence: 99%

Assessment of Soil Sustainability Using the LUCAS Database in the Southwest Region of Romania

et al. 2023

View full text Add to dashboard Cite

To ensure soil sustainability, the European Union considers the mitigation of the ecological, social and economic impacts and the prevention of soil degradation, which is the primary source of the ecosystem. In this respect, Land Use and Coverage Area Frame Survey (LUCAS) studies aim to investigate land use at the community level to gather information necessary for the analysis of the interactions between agriculture, environment and rural landscape and to provide estimates of agricultural areas with main crops. According to data from Eurostat, between May and October 2022, through the use of digital techniques, the levels of land coverage and land use, pastures, as well as irrigation management and structural elements in the landscape, were examined on the ground throughout the European Union. Data on the agricultural environment and soil were collected in the georeferenced points belonging to a representative sample by observing and completing the field form. At the level of the southwest region of Romania, the study was based on the inspection of 274 points by taking soil samples to analyze the quality indicators and identify key species of flowering plants. Data on land coverage and use can be used for a variety of environmental and socioeconomic projects in different fields.

show abstract

Pl@ntNet Crops: merging citizen science observations and structured survey data to improve crop recognition for agri-food-environment applications

Cited by 6 publications

References 25 publications

Accuracy assessment of land-use-land-cover maps: the semantic gap between in situ and satellite data

Accuracy assessment of land-use-land-cover maps: the semantic gap between in situ and satellite data

From simple labels to semantic image segmentation: Leveraging citizen science plant photographs for tree species mapping in drone imagery

Assessment of Soil Sustainability Using the LUCAS Database in the Southwest Region of Romania

Contact Info

Product

Resources

About