Fusion of Satellite Images and Weather Data With Transformer Networks for Downy Mildew Disease Detection

Maillet, William; Ouhami, Maryam; Hafiane, Adel

doi:10.1109/access.2023.3237082

Cited by 7 publications

(11 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Crop diseases, considered as an abnormal condition affecting growth, yield and quality of a plant, have been the subject of many diagnostic and detection approaches. In recent years, methodologies based on machine learning and deep learning have been developed to detect plant diseases [14], [15], [6], [7], [16].…”

Section: Detection Of Crop Diseasesmentioning

confidence: 99%

“…This model allows to focus on all parts of the input image according to their importance for the current task, thus capturing long-term dependencies in this data. ViT has demonstrated impressive performance in recent years in many computer vision tasks, especially in leaf disease detection and classification [19], [6], [20], [7]. ViT offers the ability to process high resolution images with high levels of detail, learn global context, and be pre-trained on large amounts of data.…”

Section: Detection Of Crop Diseasesmentioning

confidence: 99%

“…According to the results obtained in comparison with AlexNet and GoogleNet transfer learning approaches, attention blocks increase accuracy, but their combination with CNN blocks compensates for the speed of disease prediction. A new framework using heterogeneous data fusion and Transformer-like networks for late blight disease detection in vineyards is proposed in [7]. The authors exploit the complementarity of heterogeneous data (satellite images and weather data) by proposing a method based on three Transformers for data fusion.…”

Section: Detection Of Crop Diseasesmentioning

confidence: 99%

“…Advances in these technologies in the field of agriculture have opened up promising opportunities for early detection and diagnosis of crop abnormalities [5]. Deep neural networks, such as convolutional neural networks (CNNs) or recurrent neural networks (RNNs), as well as modern approaches based on attention mechanisms (e.g., Vision Transformers) [6], [7], are widely used, and have demonstrated outstanding performance in crop anomaly detection and diagnosis [8], [9], [10].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Image-based crop disease detection with federated learning

KABALA,

HAFIANE,

BOBELIN

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

Crop disease detection and management is critical to improving productivity, reducing costs, and promoting environmentally friendly crop treatment methods. Modern technologies, such as data mining and machine learning algorithms, have been used to develop automated crop disease detection systems. However, centralized approach to data collection and model training induces challenges in terms of data privacy, availability, and transfer costs. To address these challenges, federated learning appears to be a promising solution. In this paper, we explored the application of federated learning for crop disease classification using image analysis. We developed and studied convolutional neural network (CNN) models and those based on attention mechanisms, in this case vision transformers (ViT), using federated learning, leveraging an open access image dataset from the "PlantVillage" platform. Experiments conducted concluded that the performance of models trained by federated learning is influenced by the number of learners involved, the number of communication rounds, the number of local iterations and the quality of the data. With the objective of highlighting the potential of federated learning in crop disease classification, among the CNN models tested, ResNet50 performed better in several experiments than the other models, and proved to be an optimal choice, but also the most suitable for a federated learning scenario. The ViT_B16 and ViT_B32 Vision Transformers require more computational time, making them less suitable in a federated learning scenario, where computational time and communication costs are key parameters. The paper provides a state-of-the-art analysis, presents our methodology and experimental results, and concludes with ideas and future directions for our research on using federated learning in the context of crop disease classification.

show abstract

Section: Detection Of Crop Diseasesmentioning

confidence: 99%

Section: Detection Of Crop Diseasesmentioning

confidence: 99%

Section: Detection Of Crop Diseasesmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Image-based crop disease detection with federated learning

KABALA,

HAFIANE,

BOBELIN

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Advances in these technologies in the field of agriculture have opened up promising opportunities for early detection and diagnosis of crop abnormalities 5 . Deep neural networks, such as convolutional neural networks (CNNs) or recurrent neural networks (RNNs), as well as modern approaches based on attention mechanisms (e.g., Vision Transformers) 6 , 7 , are widely used, and have demonstrated outstanding performance in crop anomaly detection and diagnosis 8 – 10 .…”

Section: Introductionmentioning

confidence: 99%

Image-based crop disease detection with federated learning

Mamba Kabala,

Hafiane,

Bobelin

et al. 2023

Sci Rep

Self Cite

View full text Add to dashboard Cite

Crop disease detection and management is critical to improving productivity, reducing costs, and promoting environmentally friendly crop treatment methods. Modern technologies, such as data mining and machine learning algorithms, have been used to develop automated crop disease detection systems. However, centralized approach to data collection and model training induces challenges in terms of data privacy, availability, and transfer costs. To address these challenges, federated learning appears to be a promising solution. In this paper, we explored the application of federated learning for crop disease classification using image analysis. We developed and studied convolutional neural network (CNN) models and those based on attention mechanisms, in this case vision transformers (ViT), using federated learning, leveraging an open access image dataset from the “PlantVillage” platform. Experiments conducted concluded that the performance of models trained by federated learning is influenced by the number of learners involved, the number of communication rounds, the number of local iterations and the quality of the data. With the objective of highlighting the potential of federated learning in crop disease classification, among the CNN models tested, ResNet50 performed better in several experiments than the other models, and proved to be an optimal choice, but also the most suitable for a federated learning scenario. The ViT_B16 and ViT_B32 Vision Transformers require more computational time, making them less suitable in a federated learning scenario, where computational time and communication costs are key parameters. The paper provides a state-of-the-art analysis, presents our methodology and experimental results, and concludes with ideas and future directions for our research on using federated learning in the context of crop disease classification.

show abstract