Cloud architecture for digital phenotyping and automation
Olivier Debauche, Saïd Mahmoudi, Pierre Manneback, Mathieu Massinon, Nassima Tadrist, Frédéric Lebeau, Sidi Ahmed Mahmoudi · 2017
Digital phenotyping presents a very important tool for scientists to measure with high accuracy the effects of external phenomena on plant development. Plant phenotyping is mainly based on imaging techniques. However, the number of images and parameters used to store and treat these parameters are continuously growing. Consequently, the high-throughput of data and the need of specific treatment in real or near real-time requires a large quantity of resources. Moreover, the increasing amount of particular phenotyping case studies needs the development of specific application. Cloud architectures offers means to store a wide range of numerous data and host a large quantity of specific software to process these data. In this paper, we propose a new approach that shows how logic synthesis works to match digital phenotyping need and cloud possibilities in a lambda cloud architecture in order to store and treat this important amount of data in real time. We also suggest a data platform allowing to host applications and access to the stored data within the lambda architecture. The present application platform allows to use several frameworks with a fine-grained resource use of the cluster. Finally, we develop a case study in a controlled environment system (growth chamber) where we grow basil plants.