Post-Doc: image and IoT data analysis with deep learning for crop disease detection.
Keywords: multispectral imaging, remote sensing, IoT, image processing, deep learning, data fusion, digital agriculture.
Context: crop diseases cause economic losses, reduces quality and yields, and impacts negatively the environment when intensive chemical products are used for treatment. The detection of crop diseases is therefore a major challenge for agriculture, but also for the economy and the environment. Modern technologies, such as IoT, drones, remote sensing, big data and artificial intelligence, integrated by information and communication technologies, have opened a new era for digital agriculture. Indeed, these technologies offer enormous potential to solve challenging problems such as early detection and management of diseases. The state of the art shows a trend towards their widespread implementation. On the other hand, these advances have raised many challenges, such as the processing and analysis of heterogeneous data, the reliability of models and their generalization, etc.
Objective: crop disease monitoring can be performed using environmental and vegetation cover data, obtained from multispectral cameras, and IoT equipment. Promising approaches have been proposed, in recent years, using deep learning (convolutional networks, recurrent networks, transformers, ...) and data from different types of sensors. However, the heterogeneous data (sensors: image, weather, etc.) weakly labeled, make it difficult to build effective models. The aim is to develop methods based on deep network principles using recent paradigms such as self-supervised learning, attention mechanism, ... to detect and predict grapevine diseases.
Profile: PhD degree with experience in machine learning for analysis of images and data with different modalities.
Duration: ~18 months
Applications: CV to be sent to firstname.lastname@example.org; email@example.com
(c) GdR 720 ISIS - CNRS - 2011-2022.