Yawogan Jean Eudes Gbodjo
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Yawogan Jean Eudes Gbodjo has a total of 33 co-authors. Here are the top co-authors by number of shared publications:
- Dino Ienco
INRAE, UMR TETIS, University of Montpellier, Montpellier, France
9 shared publications - Louise Leroux
CIRAD, UPR AIDA, Nairobi, Kenya
4 shared publications - Raffaele Gaetano
CIRAD, Centre de Coopération Internationale en Recherche Agronomique pour le Développement, UMR TETIS, Montpellier, Paris, France
3 shared publications - Roberto Interdonato
INRIA, University of Montpellier
3 shared publications A. Bégué
1 shared publicationA.A. Balde
1 shared publication
Latest Publications
- Open Access
- 6 June 2026
- Published by Elsevier in SSRN Electronic Journal
Assessing the generalization of geospatial embeddings for temporaltransfer in land cover classification
Nowadays, remote sensing through Earth observation (EO) satellites is the preferred means of mapping land cover at various scales, from the regional to global level. While obtaining satellite data on most surface areas is more and more democratized as there are several publicly available sources, the ground truth (GT) data gap still prevents the frequent update of land cover maps or the production of historical maps in certain regions, notably in Sub-Saharan Africa. It is common for GT to be available only for a few specific years over study sites, and directly applying a model learned on a particular year to other mapping years often fails because of the data distribution shifts caused by different acquisition conditions and environmental or climatic factors. Unsupervised domain adaptation (UDA), a specific case of transfer learning, has been examined in recent years to address data distribution shifts, allowing the temporal transfer in land cover mapping of a model learned in a specific mapping year with GT data available to other mapping years without GT, using satellite image time series (SITS). In the meantime, the availability of massive unlabeled EO data and recent advances in self-supervised learning have enabled the development of geospatial foundation models and the release of geospatial embeddings such as those of AlphaEarth Foundations (AEF) and TESSERA. Although they offer a practical way to temporal transfer as they are devised to encode temporal dynamics and are fixed year-to-year vector representations, their potential for temporal transfer remains under explored. In this work, we proposed to assess their generalization for temporal transfer scenarios in land cover mapping, using a multi-year open dataset collected in Koumbia, Burkina Faso. First, we evaluated the AEF and TESSERA embeddings through a direct temporal transfer approach, and then through the UDA paradigm. Our results notably highlight the better fit of geospatial embeddings on Sentinel-2 SITS for temporal transfer tasks and the competitive performances obtained with TESSERA with respect to AEF embeddings.
- 6 June 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessAssessing the generalization of geospatial embeddings for temporaltransfer in land cover classification
Nowadays, remote sensing through Earth observation (EO) satellites is the preferred means of mapping land cover at various scales, from the regional to global level. While obtaining satellite data on most surface areas is more and more democratized as there are several publicly available sources, the ground truth (GT) data gap still prevents the frequent update of land cover maps or the production of historical maps in certain regions, notably in Sub-Saharan Africa. It is common for GT to be available only for a few specific years over study sites, and directly applying a model learned on a particular year to other mapping years often fails because of the data distribution shifts caused by different acquisition conditions and environmental or climatic factors. Unsupervised domain adaptation (UDA), a specific case of transfer learning, has been examined in recent years to address data distribution shifts, allowing the temporal transfer in land cover mapping of a model learned in a specific mapping year with GT data available to other mapping years without GT, using satellite image time series (SITS). In the meantime, the availability of massive unlabeled EO data and recent advances in self-supervised learning have enabled the development of geospatial foundation models and the release of geospatial embeddings such as those of AlphaEarth Foundations (AEF) and TESSERA. Although they offer a practical way to temporal transfer as they are devised to encode temporal dynamics and are fixed year-to-year vector representations, their potential for temporal transfer remains under explored. In this work, we proposed to assess their generalization for temporal transfer scenarios in land cover mapping, using a multi-year open dataset collected in Koumbia, Burkina Faso. First, we evaluated the AEF and TESSERA embeddings through a direct temporal transfer approach, and then through the UDA paradigm. Our results notably highlight the better fit of geospatial embeddings on Sentinel-2 SITS for temporal transfer tasks and the competitive performances obtained with TESSERA with respect to AEF embeddings.
- 6 April 2026
- Published by Agence Bibliographique de l'Enseignement Supérieur
Amélioration des systèmes de suivi des cultures à à l’aide de la télédétection multi-source et des techniques d’apprentissage profond
Les systèmes de suivi des cultures jouent un rôle essentiel dans l'évaluation de la production agricole dans le monde. De nos jours, la disponibilité de plusieurs sources d'information satellitaire à large échelle, à haute résolution spatiale et à forte répétitivité temporelle, conjointe à l'essor des techniques d'apprentissage profond, offrent de nouvelles perspectives aux systèmes de suivi des cultures pour l'évaluation de la production agricole. Dans cette thèse, nous explorons des pistes méthodologiques pour améliorer le suivi de la production agricole à partir de la télédétection multi-source et des techniques d'apprentissage profond. Nous proposons deux méthodes pour caractériser l'occupation du sol et identifier les surfaces cultivées. La première approche est basée sur des réseaux de neurones récurrents équipés de mécanismes d'attention, employant des séries temporelles multi-sources radar et optique ainsi que des connaissances spécifiques de domaine. La seconde approche repose sur des réseaux de neurones convolutifs et explore davantage la combinaison multi-source et surtout multi-échelle grâce à l'intégration d'une source optique à très haute résolution spatiale. Nous évaluons ces méthodes à des échelles territoriale et locale en ayant systématiquement un regard croisé sur des sites d'études contrastés en agriculture conventionnelle et petite agriculture familiale. Nous menons également un travail d'investigation sur l'estimation et la prévision des rendements des surfaces cultivées, à l'échelle locale de la petite agriculture familiale en employant des séries temporelles multi-sources radar et optique. Dans ce contexte en outre limité par la disponibilité de données de référence, nous évaluons le potentiel de méthodes d'apprentissage profond par rapport à des approches traditionnellement utilisées. Globalement, l'évaluation des approches proposées pour identifier les surfaces cultivées montre que les techniques d'apprentissage profond semblent mieux adaptées que les méthodes traditionnelles d'apprentissage automatique pour tirer parti de la complémentarité des données multi-sources, multi-temporelles et multi-échelles à mesure qu'il y ait une quantité suffisante de données pour leur entraînement supervisé. Le travail d'investigation réalisé pour l'estimation et la prévision des rendements n'a par contre pas révélé de plus-value manifeste dans l'emploi de ces méthodes. Dans ce dernier cas, le contexte limité en données d'entraînement semble en être la principale explication.
- 6 April 2026
- Published by Agence Bibliographique de l'Enseignement Supérieur
Amélioration des systèmes de suivi des cultures à à l’aide de la télédétection multi-source et des techniques d’apprentissage profond
Les systèmes de suivi des cultures jouent un rôle essentiel dans l'évaluation de la production agricole dans le monde. De nos jours, la disponibilité de plusieurs sources d'information satellitaire à large échelle, à haute résolution spatiale et à forte répétitivité temporelle, conjointe à l'essor des techniques d'apprentissage profond, offrent de nouvelles perspectives aux systèmes de suivi des cultures pour l'évaluation de la production agricole. Dans cette thèse, nous explorons des pistes méthodologiques pour améliorer le suivi de la production agricole à partir de la télédétection multi-source et des techniques d'apprentissage profond. Nous proposons deux méthodes pour caractériser l'occupation du sol et identifier les surfaces cultivées. La première approche est basée sur des réseaux de neurones récurrents équipés de mécanismes d'attention, employant des séries temporelles multi-sources radar et optique ainsi que des connaissances spécifiques de domaine. La seconde approche repose sur des réseaux de neurones convolutifs et explore davantage la combinaison multi-source et surtout multi-échelle grâce à l'intégration d'une source optique à très haute résolution spatiale. Nous évaluons ces méthodes à des échelles territoriale et locale en ayant systématiquement un regard croisé sur des sites d'études contrastés en agriculture conventionnelle et petite agriculture familiale. Nous menons également un travail d'investigation sur l'estimation et la prévision des rendements des surfaces cultivées, à l'échelle locale de la petite agriculture familiale en employant des séries temporelles multi-sources radar et optique. Dans ce contexte en outre limité par la disponibilité de données de référence, nous évaluons le potentiel de méthodes d'apprentissage profond par rapport à des approches traditionnellement utilisées. Globalement, l'évaluation des approches proposées pour identifier les surfaces cultivées montre que les techniques d'apprentissage profond semblent mieux adaptées que les méthodes traditionnelles d'apprentissage automatique pour tirer parti de la complémentarité des données multi-sources, multi-temporelles et multi-échelles à mesure qu'il y ait une quantité suffisante de données pour leur entraînement supervisé. Le travail d'investigation réalisé pour l'estimation et la prévision des rendements n'a par contre pas révélé de plus-value manifeste dans l'emploi de ces méthodes. Dans ce dernier cas, le contexte limité en données d'entraînement semble en être la principale explication.
- Open Access
- 25 December 2025
- Published by Elsevier in Remote Sensing of Environment
- Vol. 334
Self-supervised representation learning for cloud detection using Sentinel-2 images
- 25 December 2025
- Published by Elsevier in Remote Sensing of Environment
- Vol. 334
Open AccessSelf-supervised representation learning for cloud detection using Sentinel-2 images
- 23 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 22 (1545598X) , 1-5
Leveraging Physical Augmentations From Multiview Remote Sensing Images for Building Segmentation
Building segmentation from remote sensing data has been boosted in recent years by the advances in data processing algorithms and the improved ability of sensors to capture very high spatial resolution images. The prevailing approaches to this task are built upon deep semantic segmentation networks learned on ortho geometry i.e. orthorectified images. In this study, we propose to deal with the building segmentation using multi-view Pléiades satellite images at the perfect sensor geometry instead of the ortho geometry. This has two main advantages: (i) it frees the segmentation workflow from geometric imprecision that may arise after the image rectification step and (ii) it allows the physical augmentations of the ground truth (GT) by re-projecting building objects on each of the multi-view acquisitions. The GT re-projection process makes use of Rational Polynomial Coefficients (RPC) provided as image metadata and a fine scale Digital Surface Model. We assess the benefit of our proposal using a U-Net encoder-decoder learned on a dataset composed of tri-stereo Pléiades acquisitions over 6 french cities. Experimental results demonstrate the significance of the proposal especially an enhanced generalization capability for the building segmentation.
- 23 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 22 (1545598X) , 1-5
Leveraging Physical Augmentations From Multiview Remote Sensing Images for Building Segmentation
Building segmentation from remote sensing data has been boosted in recent years by the advances in data processing algorithms and the improved ability of sensors to capture very high spatial resolution images. The prevailing approaches to this task are built upon deep semantic segmentation networks learned on ortho geometry i.e. orthorectified images. In this study, we propose to deal with the building segmentation using multi-view Pléiades satellite images at the perfect sensor geometry instead of the ortho geometry. This has two main advantages: (i) it frees the segmentation workflow from geometric imprecision that may arise after the image rectification step and (ii) it allows the physical augmentations of the ground truth (GT) by re-projecting building objects on each of the multi-view acquisitions. The GT re-projection process makes use of Rational Polynomial Coefficients (RPC) provided as image metadata and a fine scale Digital Surface Model. We assess the benefit of our proposal using a U-Net encoder-decoder learned on a dataset composed of tri-stereo Pléiades acquisitions over 6 french cities. Experimental results demonstrate the significance of the proposal especially an enhanced generalization capability for the building segmentation.
- 14 June 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. abs 1806 1, 621-626
Dealing with missing modalities at test time for land cover mapping: A case study on multi-source optical data
In recent years, multiple sources of remote sensing data have become increasingly available to monitor Earth’s surface phenomena. However, unlike High Spatial Resolution (HSR) data, Very High Spatial Resolution (VHSR) satellite images remain difficult to collect over large areas due to acquisition costs and a smaller swath. This often compromises the simultaneous use of both sources of data over same study areas for many applications. In this work, we investigate a land cover mapping setting in which both HSR and VHSR are available at the learning stage of a deep neural network while only the HSR data is available at inference time for model inference. We thus propose simple but effective strategies for enhancing the land cover classification in this scenario of incomplete multi-source remote sensing data when the model is deployed.
- 14 June 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. abs 1806 1, 621-626
Dealing with missing modalities at test time for land cover mapping: A case study on multi-source optical data
In recent years, multiple sources of remote sensing data have become increasingly available to monitor Earth’s surface phenomena. However, unlike High Spatial Resolution (HSR) data, Very High Spatial Resolution (VHSR) satellite images remain difficult to collect over large areas due to acquisition costs and a smaller swath. This often compromises the simultaneous use of both sources of data over same study areas for many applications. In this work, we investigate a land cover mapping setting in which both HSR and VHSR are available at the learning stage of a deep neural network while only the HSR data is available at inference time for model inference. We thus propose simple but effective strategies for enhancing the land cover classification in this scenario of incomplete multi-source remote sensing data when the model is deployed.
- 16 December 2021
- Published by Taylor & Francis in Remote Sensing Letters
- Vol. 13 (2) , 162-172
Combining pixel- and object-level information for land-cover mapping using time-series of Sentinel-2 satellite data
In this letter, we propose a new methodology for Satellite Image Time Series (SITS) land cover mapping, named Two Branches Convolutional Neural Network (TwoBCNN). The main objective of the proposed methodology is to combine pixel- and object-level multi-variate time-series information in the classification process. Experiments were carried out on a study site located in the south-west of France, namely, Dordogne leveraging Sentinel-2 SITS data. Results are compared to those obtained by several standard used approaches to deal with SITS-based land cover mapping. Results demonstrate that TwoBCNN, based on a combination of pixel- and object-based information, achieved the highest classification performance with respect to the competing approaches.
- 16 December 2021
- Published by Taylor & Francis in Remote Sensing Letters
- Vol. 13 (2) , 162-172
Combining pixel- and object-level information for land-cover mapping using time-series of Sentinel-2 satellite data
In this letter, we propose a new methodology for Satellite Image Time Series (SITS) land cover mapping, named Two Branches Convolutional Neural Network (TwoBCNN). The main objective of the proposed methodology is to combine pixel- and object-level multi-variate time-series information in the classification process. Experiments were carried out on a study site located in the south-west of France, namely, Dordogne leveraging Sentinel-2 SITS data. Results are compared to those obtained by several standard used approaches to deal with SITS-based land cover mapping. Results demonstrate that TwoBCNN, based on a combination of pixel- and object-based information, achieved the highest classification performance with respect to the competing approaches.
- Open Access
- 9 November 2021
- Published by Taylor & Francis in International Journal of Remote Sensing
- Vol. 42 (24) , 9285-9308
Benchmarking statistical modelling approaches with multi-source remote sensing data for millet yield monitoring: a case study of the groundnut basin in central Senegal
In Sub-Saharan Africa, smallholder farms play a key role in agriculture, occupying most of the agricultural land. Design policies for increasing smallholder productivity remains a safe way to establish sustainable food systems and boost local economies. However, efforts are still needed in order to achieve accurate and timely monitoring in smallholder farming systems. With the advent of modern Earth Observation programmes such as the Sentinel satellites, which provide quasi-synchronous and high-resolution multi-source information over any area of the continental surfaces, new opportunities are opened up to accurately map crop yields in smallholder farming systems. This study intends to estimate and forecast millet yields in central Senegal, making the use of multi-source (synthetic-aperture radar (SAR) and optical) image time series and state-of-the-art machine learning models. A Random Forest (RF) model explained up to 50% of the millet yield variability, while deep learning models such as Convolutional Neural Network (CNN) showed promise results but performed lower. We also found that the concatenation of SAR polarizations and vegetation indices improved our crop yield modelling, but such improvement was tightly related to the modelling approach, namely RF and CNN. Using RF to forecast millet yields, we achieved stable and satisfactory accuracy 2 weeks before the harvest period.
- 9 November 2021
- Published by Taylor & Francis in International Journal of Remote Sensing
- Vol. 42 (24) , 9285-9308
Open AccessBenchmarking statistical modelling approaches with multi-source remote sensing data for millet yield monitoring: a case study of the groundnut basin in central Senegal
In Sub-Saharan Africa, smallholder farms play a key role in agriculture, occupying most of the agricultural land. Design policies for increasing smallholder productivity remains a safe way to establish sustainable food systems and boost local economies. However, efforts are still needed in order to achieve accurate and timely monitoring in smallholder farming systems. With the advent of modern Earth Observation programmes such as the Sentinel satellites, which provide quasi-synchronous and high-resolution multi-source information over any area of the continental surfaces, new opportunities are opened up to accurately map crop yields in smallholder farming systems. This study intends to estimate and forecast millet yields in central Senegal, making the use of multi-source (synthetic-aperture radar (SAR) and optical) image time series and state-of-the-art machine learning models. A Random Forest (RF) model explained up to 50% of the millet yield variability, while deep learning models such as Convolutional Neural Network (CNN) showed promise results but performed lower. We also found that the concatenation of SAR polarizations and vegetation indices improved our crop yield modelling, but such improvement was tightly related to the modelling approach, namely RF and CNN. Using RF to forecast millet yields, we achieved stable and satisfactory accuracy 2 weeks before the harvest period.
- Open Access
- 11 October 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 14 (19391404) , 11485-11499
Multisensor Land Cover Classification With Sparsely Annotated Data Based on Convolutional Neural Networks and Self-Distillation
Extensive research studies have been conductedin recent years to exploit the complementarity among multi-sensor (or multi-modal) remote sensing data for prominentapplications such as land cover mapping. In order to make a step further with respect to previous studies which investigate multi-temporal SAR and optical data or multi-temporal/multi-scale optical combinations, here we propose a deep learning framework that simultaneously combine all these input sources, specifically multi-temporal SAR/optical data and fine scale optical information. Our proposal relies on a patch-based multi-branch convolutional neural network (CNN) that exploits different per source CNN encoders to deal with the specificity of the input signals. Additionally, our framework is equipped with a self-distillation strategy to boost the per source analyses. This new strategy leverages the final prediction of the multi-source framework to guide the learning of the per source CNN encoders supporting the network to learn from itself.Experiments are carried out on two real world benchmarks, namely the Reunion island (a french overseas department) and the Dordogne study site (a southwest department in France)where the annotated reference data was collected under operational constraints (sparsely annotated ground truth data). Obtained results, providing an overall classification accuracyof about 94% (resp. 88%) on theReunion island(resp. theDordogne) study site. These findings highlight the effectivenessof our framework based on Convolutional Neural Networks andself-distillation to combine heterogeneous multi-sensor remotesensing data and confirm the benefit of multi-modal analysisfor downstream tasks such as land cover mapping.
- 11 October 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 14 (19391404) , 11485-11499
Open AccessMultisensor Land Cover Classification With Sparsely Annotated Data Based on Convolutional Neural Networks and Self-Distillation
Extensive research studies have been conductedin recent years to exploit the complementarity among multi-sensor (or multi-modal) remote sensing data for prominentapplications such as land cover mapping. In order to make a step further with respect to previous studies which investigate multi-temporal SAR and optical data or multi-temporal/multi-scale optical combinations, here we propose a deep learning framework that simultaneously combine all these input sources, specifically multi-temporal SAR/optical data and fine scale optical information. Our proposal relies on a patch-based multi-branch convolutional neural network (CNN) that exploits different per source CNN encoders to deal with the specificity of the input signals. Additionally, our framework is equipped with a self-distillation strategy to boost the per source analyses. This new strategy leverages the final prediction of the multi-source framework to guide the learning of the per source CNN encoders supporting the network to learn from itself.Experiments are carried out on two real world benchmarks, namely the Reunion island (a french overseas department) and the Dordogne study site (a southwest department in France)where the annotated reference data was collected under operational constraints (sparsely annotated ground truth data). Obtained results, providing an overall classification accuracyof about 94% (resp. 88%) on theReunion island(resp. theDordogne) study site. These findings highlight the effectivenessof our framework based on Convolutional Neural Networks andself-distillation to combine heterogeneous multi-sensor remotesensing data and confirm the benefit of multi-modal analysisfor downstream tasks such as land cover mapping.
- Open Access
- 29 January 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 9, 23070-23082
Attentive Spatial Temporal Graph CNN for Land Cover Mapping From Multi Temporal Remote Sensing Data
Satellite image time series (SITS) collected by modern Earth Observation (EO) systems represent a valuable source of information that supports several tasks related to the monitoring of the Earth surface dynamics over large areas. A main challenge is then to design methods able to leverage the complementarity between the temporal dynamics and the spatial patterns that characterize these data structures. Focusing on land cover classification (or mapping) tasks, the majority of approaches dealing with SITS data only considers the temporal dimension, while the integration of the spatial context is frequently neglected. In this work, we propose an attentive spatial temporal graph convolutional neural network that exploits both spatial and temporal dimensions in SITS. Despite the fact that this neural network model is well suited to deal with spatio-temporal information, this is the first work that considers it for the analysis of SITS data. Experiments are conducted on two study areas characterized by different land cover landscapes and real-world operational constraints (i.e., limited labeled data due to acquisition costs). The results show that our model consistently outperforms all the competing methods obtaining a performance gain, in terms of F-Measure, of at least 5 points with respect to the best competing approaches on both benchmarks.
- 29 January 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 9, 23070-23082
Open AccessAttentive Spatial Temporal Graph CNN for Land Cover Mapping From Multi Temporal Remote Sensing Data
Satellite image time series (SITS) collected by modern Earth Observation (EO) systems represent a valuable source of information that supports several tasks related to the monitoring of the Earth surface dynamics over large areas. A main challenge is then to design methods able to leverage the complementarity between the temporal dynamics and the spatial patterns that characterize these data structures. Focusing on land cover classification (or mapping) tasks, the majority of approaches dealing with SITS data only considers the temporal dimension, while the integration of the spatial context is frequently neglected. In this work, we propose an attentive spatial temporal graph convolutional neural network that exploits both spatial and temporal dimensions in SITS. Despite the fact that this neural network model is well suited to deal with spatio-temporal information, this is the first work that considers it for the analysis of SITS data. Experiments are conducted on two study areas characterized by different land cover landscapes and real-world operational constraints (i.e., limited labeled data due to acquisition costs). The results show that our model consistently outperforms all the competing methods obtaining a performance gain, in terms of F-Measure, of at least 5 points with respect to the best competing approaches on both benchmarks.
- 1 September 2020
- Published by Elsevier in Agricultural Systems
- Vol. 184
Using remote sensing to assess the effect of trees on millet yield in complex parklands of Central Senegal
- 1 September 2020
- Published by Elsevier in Agricultural Systems
- Vol. 184
Using remote sensing to assess the effect of trees on millet yield in complex parklands of Central Senegal