Skip to main content



Reporting period: 2019-01-01 to 2020-06-30

According to McKinsey & Company, about a third of food produced is lost or wasted every year, at the same time when 50% more and better food will be needed over the next 20-30 years. Inefficiencies in planting, harvesting, water use, reduced animal contributions, as well as uncertainty about weather, pests, consumer demand, supply chain logistics, food recalls and other intangibles contribute to the loss. Global food waste and loss cost $940 billion a year, has a carbon footprint of 4.4 Gt CO2-equivalent (more than 8 percent of global greenhouse-gas emissions), and a blue-water footprint of about 250 cubic km (3.6 times the annual consumption of the US). These challenges are being addressed in part through the development of Precision Agriculture and Precision Livestock Farming (PA/PFL). Such approaches seek to harness the power of pervasive data and technologies to improve decision making in food production.
Unfortunately, there still remains major technical challenges in empowering farmers and food producers to realise the full potential of PA/PLF. Primarily the in-accessibility of High Performance Computing platforms that are required to process large and complex data sets.
The CYBELE project aims to foster the use of HPC in PA/PLF applications and will do so by developing a suite of tools that make it much easier and faster for farmers, food producers and related stakeholders to build applications that can be rapidly deployed to HPCs, that can take advantage of pre-defined Airtifical Intelligence powered algorithms to improve decision making, which in turn will improve food quality and yield, and ultimately reduce food waste. Through the close collaboration of 9 demonstrators, the CYBELE project will deliver a state of the art data analytics suite of tools that can be used in production to make HPC more accessible and to improve time to market of new products and services that leverage AI for improved decision making.
The objectives of CYBELE are to:
* Demonstrate the value of CYBELE component in real world settings;
* Promote the technologies and solutions to the AgriFood market, in particular PA/PLF;
* Effectively exploit the results of CYBELE by creating new products and new services;
* Build tools that make HPC systems more accessible;
* Ensure data can be processed securely and appropriately for PA/PFL applications.
The project has produced 29 deliverables covering the primary use cases and requirement that the CYBELE tools should cover. These requirements were validated by industry and translated into development priorities to be designed, developed and deployed as part of the CYBELE first deployment. This first deployment encompassed HPC / Big Data abstractions layers, secure service access (WP2), the semantic processing or diverse data sets (WP3), the design of bespoke Machine Learning and Deep Learning algorithms targeted for use in PA/PLF applications (WP4), the continuous integrations of all these components into leading HPC centres across Europe (WP5). Early work has already begun on the use of the first deployment of the CYBELE components into the 9 demonstrators and client sites across the EU (WP6 & 7). CYBELE has had many early successes which have been promotes at leading international conferences and published in leading international journals (e.g. Computer Networks) (WP8), and we have invested time into carefully planning the technological and business impact of the new technological advancements in the AgriFood markets in the EU (WP9).
Major highlights include the deployment of all software artefacts to all four of the HPC centres involved in the project which represented a coordinated effort across WP2, WP3, WP4 and WP5, the technical WPs in CYBELE.
During the reporting period, CYBELE partners have pursued diverse research priorities and topics to extend the State of the Art in their field of expertise. These include: i) the convergence of hybrid Big Data – HPC resources in cloud infrastructures in a seamless manner; ii) the materialisation of CYBELE Data Model for interoperable services and contextually aligned applications; and iii) the training and usage of novel Artificial Neural Networks architectures in parallelized computing tasks.
More precisely, University of Copenhagen participated to the 9th European Conference on Precision Livestock, representing the CYBELE project in the subject of automating monitoring of pen fouling in growing pigs using Convolutional Neural Networks (CNN) applied to images captured above the pens. Also, CERTH presented the CYBELE Common Semantic Model at the 7th International Workshop on Semantic Statistics jointly organised with the International Semantic Web Conference (ISWC) which is the premier venue for presenting fundamental research, innovative technology, and applications concerning semantics, data, and the Web. Last but not least, the CYBELE technical partners (RYAX, HLRS, PSNC, ICCS, LEANSCALE, and UBITECH) presented the paper “Converging HPC, Big Data and Cloud technologies for precision agriculture data analytics on supercomputers” at the 15th Workshop on Virtualization in High-Performance Cloud Computing emphasizing on the ways to improve the co-execution of mixed HPC and Big Data workloads within heterogeneous computing facilities.
Below is a List of Publications extending SotA in the frame of CYBELE
Jensen, D. B., Larsen, M. L. V., & Pedersen, L. J. (2019), Comparison of architectures and training strategies for convolutional neural networks intended for location-specific counting of slaughter pigs. 9th European Conference on Precision Livestock Farming.
Konstantinos Perakis, Fenareti Lampathaki, Konstantinos Nikas, Yiannis Georgiou, Oskar Marko, Jarissa Maselyne. (2020). CYBELE – Fostering precision agriculture & livestock farming through secure access to large-scale HPC enabled virtual industrial experimentation environments fostering scalable big data analytics. Computer Networks, Volume 168, 2020, 107035.
Spiros Mouzakitis, Giannis Tsapelas, Sotiris Pelekis, Simos Ntanopoulos, Dimitris Askounis, Sjoukje Osing, Ioannis N.Athanasiadis (2020). Investigation of common big data analytics and decision-making requirements across diverse precision agriculture and livestock farming use cases. Proceedings of 13th International Symposium on Environmental Software Systems, February 5-7, 2020, Wageningen, The Netherlands.
Samson Damilola Fabiyi, Hai Vu, Christos Tachtatzis, Paul Murray, David Harle, Trung Kien Dao, Ivan Andonovic, Jinchang Ren, Stephen Marshall. (2020). Varietal Classification of Rice Seeds Using RGB and Hyperspectral Images. IEEE Access, vol. 8, pp. 22493-22505, 2020
Yiannis Georgiou, Naweiluo Zhou, Li Zhong, Dennis Hoppe, Marcin Pospieszny, Nikela Papadopoulou, Kostis Nikas, Orestis Lagkas Nikolos, Pavlos Kranas, Sophia Karagiorgou, Eric Pascolo, Michael Mercier and Pedro Velho. Converging HPC, Big Data and Cloud technologies for precision agriculture data analytics on supercomputers. Proceedings of VHPC - ISC 2020.
Dimitris Zeginis, Evangelos Kalampokis, Konstantinos Tarabanis (2019).202fStatistical Challenges Towards a Semantic Model for Precision Agriculture and Precision Livestock Farming 20th International Workshop on Semantic Statistics (SemStats2019) co-located with the 18th International Semantic Web Conference (ISWC2019), CEUR-WS, Vol-2549.