Final Report Summary - PHIDIAS (PHIDIAS: Phenotyping with a High-throughput, Intelligent, Distributed, and Interactive Analysis System)
Understanding biological function and development relies on unraveling the interaction between genetic information and the environment, and how they synergistically affect the phenotype of the plants.
PHIDIAS’ overriding objective is to introduce solutions for the acquisition of phenotyping information and analysis via affordable image-based techniques. This will enable laboratories throughout the world to embark on phenotyping analysis towards an improved understanding of plant biology. PHIDIAS’s goal is to increase throughput in phenotyping in a global fashion, rather than the current local fashion where costly high throughput installations are in place.
Towards achieving this goal, in the second phase of the PHIDIAS project’s duration the following have been obtained:
• Improved sensing methodologies have been developed that utilize affordable equipment such as commercial digital cameras and Rasberry Pi PCs enabling acquisition.
• Thousands of images from Arabidopsis phenotyping experiments have been collected in house and in collaboration with other laboratories. Hundreds of such images have been annotated.
• An image analysis platform was developed; it can be executed in a standalone fashion or in a cloud infrastructure (either within commercial entities such as Amazon AWS or within phenotyping platforms such as the iPlant Collaborative).
• New software tools have been developed to enable the semi or fully automated extraction of phenotypes in plants.
• New image compression methodologies were developed to save bandwidth for images transmitted over the network when processing in a cloud-enabled infrastructure. This enables the deployment of sensors in remote and poorer parts of the world, which are characterized by inferior Internet infrastructure.
• Overall in the past 4 years, 15 scientific papers (in peer reviewed venues) and reports have appeared or are in press detailing the current methodologies and technology. A dedicated website provides information to the scientific community.
• Several presentations at conferences and symposia have been given and invited seminar talks have been presented in laboratories and centers across Italy (Pisa, Florence, Rome) and Europe (Greece, Germany, UK) and several experts have been invited to talk on phenotyping and agriculture.
• Two workshops at top computer vision conferences have been organized to disseminate the importance of phenotyping and the challenges posed when developing computer vision solutions for phenotyping.
• The fellow has established several collaborations within IMT and beyond centered in the interests of phenotyping and the analysis of biological images. Most notable are collaborations in the USA with iPlant and Europe with phenotyping centers (e.g. Juelich, Wageningen, Nottingham).
• Events for the benefit of the host and the local community have also been held.
• The fellow led a group of several students and postdocs and acted as the director of the Pattern Recognition and Image Analysis unit (PRIAn). He was responsible for its research direction and the planning of the curriculum of the PhD in image analysis. The courses taught include agricultural issues and the topic of phenotyping thus several students have been exposed to new challenges of societal importance.
• The project site is available at: http://prian.imtlucca.it/PROJECTS/PHIDIAS/phidias.html
• The framework developed is available at: http://phenotiki.com
PHIDIAS’ overriding objective is to introduce solutions for the acquisition of phenotyping information and analysis via affordable image-based techniques. This will enable laboratories throughout the world to embark on phenotyping analysis towards an improved understanding of plant biology. PHIDIAS’s goal is to increase throughput in phenotyping in a global fashion, rather than the current local fashion where costly high throughput installations are in place.
Towards achieving this goal, in the second phase of the PHIDIAS project’s duration the following have been obtained:
• Improved sensing methodologies have been developed that utilize affordable equipment such as commercial digital cameras and Rasberry Pi PCs enabling acquisition.
• Thousands of images from Arabidopsis phenotyping experiments have been collected in house and in collaboration with other laboratories. Hundreds of such images have been annotated.
• An image analysis platform was developed; it can be executed in a standalone fashion or in a cloud infrastructure (either within commercial entities such as Amazon AWS or within phenotyping platforms such as the iPlant Collaborative).
• New software tools have been developed to enable the semi or fully automated extraction of phenotypes in plants.
• New image compression methodologies were developed to save bandwidth for images transmitted over the network when processing in a cloud-enabled infrastructure. This enables the deployment of sensors in remote and poorer parts of the world, which are characterized by inferior Internet infrastructure.
• Overall in the past 4 years, 15 scientific papers (in peer reviewed venues) and reports have appeared or are in press detailing the current methodologies and technology. A dedicated website provides information to the scientific community.
• Several presentations at conferences and symposia have been given and invited seminar talks have been presented in laboratories and centers across Italy (Pisa, Florence, Rome) and Europe (Greece, Germany, UK) and several experts have been invited to talk on phenotyping and agriculture.
• Two workshops at top computer vision conferences have been organized to disseminate the importance of phenotyping and the challenges posed when developing computer vision solutions for phenotyping.
• The fellow has established several collaborations within IMT and beyond centered in the interests of phenotyping and the analysis of biological images. Most notable are collaborations in the USA with iPlant and Europe with phenotyping centers (e.g. Juelich, Wageningen, Nottingham).
• Events for the benefit of the host and the local community have also been held.
• The fellow led a group of several students and postdocs and acted as the director of the Pattern Recognition and Image Analysis unit (PRIAn). He was responsible for its research direction and the planning of the curriculum of the PhD in image analysis. The courses taught include agricultural issues and the topic of phenotyping thus several students have been exposed to new challenges of societal importance.
• The project site is available at: http://prian.imtlucca.it/PROJECTS/PHIDIAS/phidias.html
• The framework developed is available at: http://phenotiki.com