D2.15 - Overall digital service catalogue [M24 release]

OVERALL DIGITAL SERVICES CATALOGUE M24 RELEASE – 2024-12-31 Ref. Ares(2024)9286184 - 31/12/2024 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 2 Project cofunded by the European Commission within the Digital Europe Programme Dissemination Level PU Public X CO Confidential, only for members of the consortium (including the Commission Services) □ CL Classified, as referred to in Commission decision 2001/844/EC □ Deliverable number: D2.15 Deliverable name: Overall digital service catalogue [M24 release] Work package: WP2 Lead WP: POLIMI Lead Task: POLIMI Main contributors: Giulio Fontana [POLIMI], Michał Błaszczak [PSNC], Marcin Plociennik [PSNC], Raffaele Giaffreda [FBK], Davaadorj Battulga [FBK] Internal reviewers: Matteo Matteucci [POLIMI], Peter Riegler-Nurscher [JR] Other contributing partners: EV ILVO, HISPATEC, IDELE, JR, L-PIT, RGRD, RISE, UCO, UdL, UNINA 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 3 Contents Contents ............................................................................................................................................................................. 3 1. Executive summary .................................................................................................................................................... 5 2. Recap of M12 release of this Deliverable (D2.14) ..................................................................................................... 6 3. Evolution of digital services in AgrifoodTEF’s service catalogue ............................................................................... 6 4. Experience gained delivering digital services to customers .................................................................................... 10 5. Plans for integration of digital services with AgrifoodTEF’s upcoming digital infrastructure ................................. 34 6. Plans for integration of digital services with AgrifoodTEF’s upcoming dataspace .................................................. 35 7. Conclusions .............................................................................................................................................................. 37 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 4 Document Revision History Date Issue Author/Editor/Contributor Summary of main change 2024-09-30 V01 Giulio Fontana (POLIMI) First document version 2024-11-26 V02 Giulio Fontana (POLIMI) Integration of POLIMI contributions 2024-12-04 V02 Giulio Fontana (POLIMI) Integration of partner contributions 2024-12-10 V03 Giulio Fontana (POLIMI) Completion of POLIMI contributions, integration of partner contributions to Section 4 2024-12-19 V04 Giulio Fontana (POLIMI) Response to feedback from internal reviewers; final reading and fine-tuning 2024-12-20 V04.1 Giulio Fontana (POLIMI) Incorporation of last contributions 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 5 1. Executive summary This document provides a snapshot of the status of AgrifoodTEF’s digital services at the end of the second year of the project. The contents of this Deliverable are focused on the activities performed in 2024 to provide digital services to European companies. They depict a significantly more advanced situation with respect to the previous version of this document, published at the end of 2023 and briefly recalled by Section 2. The advancements of 2024 are highlighted and analysed in Section 3. The most important advancement experienced during 2024 is that while at the end of 2023 only 3 digital services had been actually provided to AgrifoodTEF customers (all three by Josephinum Research to Blickwinkel Digital Service), at the end of 2024 the number of digital services ongoing or completed is above 20. This is still a low number, but it shows how the efforts devoted in Year 1 and Year 2 to design and set up the elements needed to enable and support service provision are starting to produce the expected effects. Section 4 provides information about the digital services provided by AgrifoodTEF to customers in 2024. The added value provided by this section (when compared with a mere recapitulation of the services) consists of insights and “lessons learned” that we collected from service providers. Their publication as part of this Deliverable is aimed at helping AgrifoodTEF partners streamline their delivery of digital services in the future. Section 5 and Section 6 are dedicated to the evolution of the technical background underpinning AgrifoodTEF’s digital services. Specifically, Section 5 provides information about the digital infrastructure while Section 6 is focused on the AgrifoodTEF Dataspace. This technical background, still under development, is very important to support service offer, management and execution during the life of AgrifoodTEF, and even more for long-term selfsustainability. Section 7 concludes the document. 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 6 2. Recap of M12 release of this Deliverable (D2.14) When compared with the M12 release of this Deliverable1, the M24 version outlines a very different situation. The main evolution is in the number of digital services actually negotiated with customer companies and ongoing or concluded, which greatly increased: from 3 in 2023 to 23 in 2024 (the latter figure having been evaluated at the end of November). Notwithstanding this development, 2024 is still a transition year for AgrifoodTEF, for reasons that Section 3 will explain in detail. Therefore we expect the throughput of services to customers to increase yet again, and significantly, in 2025. D2.14, considering that the number of services delivered was very low due to the project having just started its activities, focused instead on providing an overview of the services listed in AgrifoodTEF’s catalogue at the end of 2023. Those numbers were: 88 physical services 79 digital services 17 conformity services 26 other services These numbers highlighted how, thanks to the extensive prior experience of the Consortium, AgrifoodTEF was capable of presenting to European industry a credible and complete portfolio of services already at the end of 2023. Thanks to the better state of development of AgrifoodTEF’s activities, D2.15 can now expand its analysis from services listed in the catalogue to services actually being provided to customers. An analysis of the outcomes of these action is provided by Section 4; before that, Section 3 provides a general overview of the progress occurred during 2024 in AgrifoodTEF’s offer of digital services. 3. Evolution of digital services in AgrifoodTEF’s service catalogue As a preliminary step before a more wide-ranging progress discussion, we report here the results of a new analysis of AgrifoodTEF’s catalogue, similar to the one performed for D2.14 but performed at the end of 2024. The version of the catalogue considered for the analysis was Version 1_10 of mid-October 2024, shortly before the focus of activity moved from the catalogue (which was essentially a list of services) to the much more flexible Online Catalogue integrated in AgrifoodTEF’s Web Portal. More will be said below about this important passage. The results of the analysis performed on Version 1_10 of the catalogue provided showed that at the end of 2024 the composition of the catalogue was as follows: 120 physical services [+32] 118 digital services [+39] 24 conformity services [+7] 51 other services [+25] The numbers between brackets show the increase in the number of services with respect to D2.14. 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 7 The total number of available services in Version 1_10 of the catalogue is 313, which correspond to 275 distinct services. 38 services, in fact, are “composite” services which belong to more than one service category: for instance, a service may be composed of two well-defined sub-services, one physical and one digital. It’s worth mentioning that wherever this is feasible AgrifoodTEF promotes modularisation of services, i.e. definition of independent subservices as separate entities that can be combined but can also be exploited independently, so over time we expect the occurrence of “composite” to become rarer. It is also important to point out that the separation of services into “digital”, “physical” and “conformity-related”, though convenient for analysis and for the association of services to Work Packages, is not an element used by AgrifoodTEF for service classification or selection by their users via the Online Catalogue. Such separation of services –useful in the context of this Deliverable– is also subject to a degree of arbitrarity, so its ultimate usefulness is more qualitative than quantitative. Figure 1 (below) provides a comparison between the service offer of AgrifoodTEF at the end of 2023 (catalogue Version 1_1) and at the end of 2024 (catalogue Version 1_10). 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 8 Figure 1: Comparison between services in the AgrifoodTEF catalogue at the end of 2024 vs the end of 2023. Overall, the numbers for 2024 show a notable increase (+49%) in the total number of services available in the catalogue. The main points of evolution concern digital services and particularly “other services”, i.e. those that do not fit easily in the other categories. A possible interpretation of the large increase in “other” services is that as partners become more confident in the machinery of service offer and provision in the context of AgrifoodTEF, they also feel more secure in expanding the scope and range of their services, bringing more of their internal competencies to AgrifoodTEF. 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 9 For what concerns the increase in the number of digital services, the evolution can be linked to an expansion from simpler services such as dataset provision to more complex ones such as training of machine learning algorithms, or even more specialised ones. This expansion is partly due to requests from companies, which pushed project partners to define new services that later become part of the general catalogue. Here, too, we notice an effort by partners to offer customers, in the form of AgrifoodTEF services, wider access to their technical capabilities with respect to 2023. This is a positive development since it both expands the catalogue of services and -over timeit promotes convergence and synergy. In fact, as multiple partners end up providing similar services, it opportunities emerge to compare their implementations and to align them, also opening up the possibility to identify and define “best practices” in the process. It is worth noting that the kind of “customer push” described above with respect to digital services is a major force towards the expansion and evolution of all of AgrifoodTEF’s catalogue. As companies get in contact with AgrifoodTEF services, partners gain a better understanding of the type of services that they need and look for, and are therefore able to fine-tune their offering to better match the necessities of customers. This better matching is expected, in turn, to make AgrifoodTEF’s catalogue more attractive to prospective customers, and ultimately to promote the usage of services by European companies. In the preceding part of this section we made reference to the shift from the initial list-like form of AgrifoodTEF’s service catalogue to its final form, i.e. an interactive selection tool integrated into the Web Portal. This shift started at the end of 2024 and is currently fully underway. The process of bringing services to the Online Catalogue is very important to improve the visibility and impact of the services towards prospective customers: in particular, the Online Catalogue incorporates the category-based service selection mechanism outlined in Deliverable D2.1 (“Digital infrastructure requirements and initial catalogue of services”), designed to let customers quickly find services of their own interest among the large number of services available, using criteria based on the user’s own needs. The shift from the list-like original catalogue to the interactive Online Catalogue within the Web Portal requires, however, to rethink and re-describe each one of AgrifoodTEF’s services. Stricter guidelines on service content and description are now in place, and only services that comply with these guidelines get represented on the Web Portal; a revision process involving technical and formal stages ensures the quality of service descriptions in the Online Catalogue. As a result of the migration to the Web Portal described above, all project partners are currently busy updating the descriptions of their pre-existing services to make them compliant with the new guidelines and thus suitable for publication on the Portal, and significant effort is being devoted to the review of content and form of the updated service descriptions. For many partners, this is also the occasion to fine-tune and expand their service offering. For these reasons, this Deliverable is a snapshot of a transition phase in the life of AgrifoodTEF’s service catalogue. We expect that, at M36, the next yearly revision of this Deliverable will describe a stabilised situation where the only relevant service repository is the Online Catalogue integrated into the Web Portal. 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 16 Code(s) and name(s) of the digital service(s) S00123 (AI hardware performance assessment) S00124 (AI algorithms performance assessment) Partner providing the service(s) FBK – Fondazione Bruno Kessler Name, country and brief profile of the customer [up to 5 lines] GeoInference, Italy, Startup providing AI and computer-vision based applications for yield estimation and sizing of crops. Problems that the customer needs to solve [up to 5 lines] Optimise the AI-based solution to be more efficient in conditions of edge deployments. Minimise memory, computational footprint and timing needed for executing live processing of video streams. Brief description of the service(s) (including any customisations) [up to 5 lines] Assess the software and hardware performance of Geoinference’s BioSmart solution, which measures quantity and caliber of apples based on video input. Service execution state Not started yet Ongoing Completed X (please check the appropriate box) Service onboarding and startup: shareable insights and experiences (if any) [up to 10 lines] Onboarding was mainly done through existing contacts. Service execution: shareable insights and experiences (if any) [up to 10 lines] The service has been delivered over a period of six months. The interactions with the SME have been more intense than planned at the beginning, due to the need to get accustomed with the way their software had been coded. Post-service analysis: shareable insights and experiences (if any) [up to 10 lines] The interactions with GeoInference enables us to find additional business opportunities from the Italian Node ecosystem supporting the AgrifoodTEF execution. FederUnacoma was involved in the discussion when it came to finding suitable ag-machinery producers willing to test further the solution on their machines. Revo Italia srl was identified and discussions ended up in adoption of the GeoInference solution with a first joint-sale (Rovo+GeoInference) happening in Q2 2024. 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 17 Code(s) and name(s) of the digital service(s) S00125 Data valorization and dataspace integration assessment Partner providing the service(s) FBK – Fondazione Bruno Kessler Name, country and brief profile of the customer [up to 5 lines] AISPOT, an Italian startup, has developed a solution to identify, early on, the presence of “alternaria”, a fungus disease that attacks fruit crops. Problems that the customer needs to solve [up to 5 lines] The startup has requested a fresh investigation with FBK experts into their solution's accuracy (ability to identify disease in plants). Brief description of the service(s) (including any customisations) [up to 5 lines] The main goal of the initial service was to analyze the startup available data, to create a highquality dataset to be used for finding and evaluating more accurate object detection techniques from recent advancements. Service execution state Not started yet Ongoing X Completed (please check the appropriate box) Service onboarding and startup: shareable insights and experiences (if any) [up to 10 lines] Onboarding was mainly done through existing contacts. Service execution: shareable insights and experiences (if any) [up to 10 lines] After having developed the mechanical solution, the company has made a data collection campaign of around 6.000 images, of which 3523 have been annotated. The resulting dataset has been used to train a Yolo v3 image detection algorithm. FBK has provided additional validation support by testing additional and more recent image detection algorithms. Post-service analysis: shareable insights and experiences (if any) [up to 10 lines] --- 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 18 Code(s) and name(s) of the digital service(s) S00041 - Providing dataset S00040 - Testing of weed detection Partner providing the service(s) JR – Josephinum Research Name, country and brief profile of the customer [up to 5 lines] Farm-ING Smart Farm Equipment GmbH, Austria Farm-ING is a start-up company in the smart farming sector. The company develops smart machines for crop care and provides engineering services to OEM. Problems that the customer needs to solve [up to 5 lines] Spot spraying implement called SprayING: The sprayer has an integrated AI-based camera system to detect weeds, which controls each nozzle with high frequency of up to 100 times per second. The Spot Sprayer can be used not only for herbicide application, but also for fungicides, insecticides, and fertilizers. Brief description of the service(s) (including any customisations) [up to 5 lines] Farm-ING wants to improve its models for the spot spraying system. The services help them by proving additional image data and by testing their model performance. The service is customized regarding the specific crops targeted by the system. Service execution state Not started yet Ongoing X Completed (please check the appropriate box) Service onboarding and startup: shareable insights and experiences (if any) [up to 10 lines] We had already cooperated with Farm-ING in advance. So, the process was very simple and straightforward. Service execution: shareable insights and experiences (if any) [up to 10 lines] Provision of the data set worked without any problems. The algorithm has yet to be tested. Post-service analysis: shareable insights and experiences (if any) [up to 10 lines] --- 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 19 Code(s) and name(s) of the digital service(s) S00172 - Data augmentation Partner providing the service(s) JR – Josephinum Research Name, country and brief profile of the customer [up to 5 lines] Neurinos GmbH, Germany is a company working with farmers, research institutions and food manufacturers to develop platform technology, sensor technology and artificial intelligence in the animal production chain. The current focus is on image recognition for behavioural monitoring of livestock in different contexts. Problems that the customer needs to solve [up to 5 lines] Neurinos develops advanced image recognition technologies to identify individual cows within a herd and assess and document their welfare in real time using continuous video monitoring. Additionally, a module of the solution is able to monitor sheep, with a particular focus on detecting signs of impending birth, health issues and general behaviour. The system will use comprehensive video datasets of sheep at various stages of labour to train AI models that can predict and monitor birth events and health anomalies. Brief description of the service(s) (including any customisations) [up to 5 lines] This service is jointly provided with RGRD. RGRD is proving video data from their barns to improve the models from Neurinos. JR provides a service for anonymization of the video data. Specifically, faces of people in the video are being blurred. Service execution state Not started yet Ongoing X Completed (please check the appropriate box) Service onboarding and startup: shareable insights and experiences (if any) [up to 10 lines] Nothing noteworthy. Service execution: shareable insights and experiences (if any) [up to 10 lines] In particular, there were problems when processing videos from fisheye cameras. It was not possible to achieve the robustness of facial recognition. Therefore, only the data from normal cameras is used. However, these will not be delivered until next year. There will therefore be a delay to this service and a reduced service volume. Post-service analysis: shareable insights and experiences (if any) [up to 10 lines] --- 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 20 Code(s) and name(s) of the digital service(s) S00318 - Real-Time Data Integration for Smart Irrigation Systems S00319 - Data Analysis for Smart Irrigation Systems S00320 - Optimized Irrigation Scheduling for Smart Agriculture Partner providing the service(s) HISPATEC Name, country and brief profile of the customer [up to 5 lines] Wiseconn Ibérica, Spain is a company specializing in irrigation valve control systems for micro and pivot irrigation. Their solution is based on a wireless network that allows monitoring of crucial factors in crops, such as climatic conditions and soil moisture levels. This facilitates more efficient irrigation management, optimizing water use and improving agricultural yields. Problems that the customer needs to solve [up to 5 lines] These services help agricultural companies test how to improve water efficiency through irrigation data analysis and the provision of actionable insights. These clients will gain a deeper understanding of water usage, soil conditions, and optimal irrigation schedules, leading to reduced water waste, increased crop yield, and cost savings. It also enhances decision-making with realtime data integration. Brief description of the service(s) (including any customisations) [up to 5 lines] These services allow companies to experiment with real-time data and advanced analytics to improve water management, reduce waste, and enhance crop yield. The service includes comprehensive testing and evaluation to ensure compatibility and performance, ensuring efficient irrigation practices and sustainability through insights into soil moisture, weather conditions, and plant water needs. Service execution state Not started yet x Ongoing Completed (please check the appropriate box) Service onboarding and startup: shareable insights and experiences (if any) [up to 10 lines] We have not encountered any difficulties during the process. At this stage, we are focused on identifying the most suitable location to conduct the experimentation, ensuring it maximizes visibility and dissemination impact. Service execution: shareable insights and experiences (if any) [up to 10 lines] --- Post-service analysis: shareable insights and experiences (if any) [up to 10 lines] --- 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 21 Code(s) and name(s) of the digital service(s) S00001 Hyperspectral measurements and analysis Partner providing the service(s) EV ILVO - Flanders Research Institute for Agriculture, Fisheries and Food Name, country and brief profile of the customer [up to 5 lines] Perkin Elmer develops solutions for laboratory analysis and measurements for various sectors. In their Agriculture and Food department, the company develops innovative solutions for measuring quality of grain, milk and other raw products. Their in-line, on-line and at-line analysers allow their customers to accurately monitor processes in real time to optimize production yield, reduce waste, and control final product quality and safety. Problems that the customer needs to solve [up to 5 lines] The company is developing a hyperspectral measurement device which can be used for routine analysis throughout the winemaking process. The device can be used to analyse both must and wine from vinification to bottling. Through chemometrics, quality parameters such as ethanol, pH, sugars, density, etc. are determined. In this service, the company wants to evaluate the performance of the device and the prediction model for Belgian wines. Brief description of the service(s) (including any customisations) [up to 5 lines] In this service, samples of Belgian wines are analysed with the device and results are compared with laboratory analysis results. Also, the raw spectra are acquired, pre-processed and specific chemometric statistical and machine learning prediction models are developed for potential further improvements. Service execution state Not started yet Ongoing X Completed (please check the appropriate box) Service onboarding and startup: shareable insights and experiences (if any) [up to 10 lines] We had already established contacts with the company for other equipment. Therefore, the discussions regarding this application proceeded smoothly and efficiently. Service execution: shareable insights and experiences (if any) [up to 10 lines] The service is currently still ongoing. To obtain a representative dataset that covers the entire parameter space, a large number of samples are required, which are acquired during multiple moments throughout the year. To account for annual variability, the service is conducted over several years. Post-service analysis: shareable insights and experiences (if any) [up to 10 lines] --- 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 22 Code(s) and name(s) of the digital service(s) S00006 Agri dataset Generation Partner providing the service(s) EV ILVO - Flanders Research Institute for Agriculture, Fisheries and Food Name, country and brief profile of the customer [up to 5 lines] SESVanderHave is a competitive, international player in the sugar beet industry. As sugar beet experts, they cover every step of the process, from the breeding of new varieties to the commercialization and distribution of sugar beet seeds. Problems that the customer needs to solve [up to 5 lines] Use of AI for the detection of Cercospora in sugar beets in high resolution UAV imagery. Applying existing disease detection AI-models (developed for potatoes) in sugar beet. Further optimize AI training and model for detection in sugar beet. Brief description of the service(s) (including any customisations) [up to 5 lines] Comparison between existing detection models and newly developed models. Comparison between different sensors. Service execution state Not started yet Ongoing Completed x (please check the appropriate box) Service onboarding and startup: shareable insights and experiences (if any) [up to 10 lines] Onboarding was mainly done through existing contacts. Service execution: shareable insights and experiences (if any) [up to 10 lines] Customer was delighted with the spatial detail that partner was able to provide. Annotation efficiency was also applauded. Post-service analysis: shareable insights and experiences (if any) [up to 10 lines] AI models were transferable to other crops but required some optimization to be usable. Knowledge about crop and diseases was necessary to translate and interpret model outputs. 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 23 Code(s) and name(s) of the digital service(s) S00004 - Robotic Software Framework Partner providing the service(s) EV ILVO - Flanders Research Institute for Agriculture, Fisheries and Food Name, country and brief profile of the customer [up to 5 lines] Inagro is a leading agricultural research and advisory organization in Flanders, Belgium. It provides practical, innovative solutions to farmers and horticulturists, helping them optimize production, sustainability, and profitability. Through applied research, Inagro develops and tests new techniques and technologies tailored to regional needs. They also offer training, workshops, and personalized advice to farmers. Inagro acts as a bridge between academic research and the agricultural sector, promoting sustainable and efficient farming practices. Problems that the customer needs to solve [up to 5 lines] Inagro manages experimental fields where various agricultural experiments are conducted. For their experimental in biologic production where a fixed tramline system (3m) is used, they require an electric implement carrier specifically designed for mechanical weed control. However, there is currently no solution on the market that fully meets their needs. Therefore, a custom development is necessary. For the design, they collaborate with machinery manufacturers, but several design parameters are essential to guide the development process. Brief description of the service(s) (including any customisations) [up to 5 lines] In this service, two experiments were conducted to gather data for determining crucial design parameters (motor dimensioning, battery capacity). During these experiments, two existing robot platforms were used. In the first experiment, the efficiency of an electro-hydraulic steering system was evaluated. During the second experiment, data was acquired during hoeing to determine the required power for this operation. Two different velocities and three different configurations were tested. Service execution state Not started yet Ongoing Completed x (please check the appropriate box) Service onboarding and startup: shareable insights and experiences (if any) [up to 10 lines] We already had good contacts with Inagro, including a collaboration in the Interreg CIMAT project where we developed an electrical 4-wheel drive, 4-wheel steering robot for mechanical and thermal weed control. That was the reason they contacted us for this experiment. Therefore, the onboarding process went smoothly. Service execution: shareable insights and experiences (if any) [up to 10 lines] Initially, tests were planned to be conducted at fields from Inagro, but this proved difficult to arrange due to specific field and weather conditions. Therefore, tests were conducted at EV ILVO fields and were performed very efficiently. Data was logged to a timeseries database, which was coupled to the ILVO ARTOF robot software framework. Post-service analysis: shareable insights and experiences (if any) [up to 10 lines] Having robots with specific drive systems available has proven very useful for conducting these types of tests, especially because they are equipped with our own developed software framework, ARTOF. This framework allows various machine parameters to be captured and logged with an exact link to their position (RTK-GNSS). This data can be captured under realistic and representative field conditions during actual field operations, making it particularly valuable for designers of agricultural robots or implements. To better interpret the results, also the electro-mechanical efficiency of the drive system was tested on a parking lot. From these tests it became clear that a more standardized and automated way of executing these efficiency tests would be very beneficial. 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 24 Code(s) and name(s) of the digital service(s) Design and implementation of certified smart irrigation systems Partner providing the service(s) UNINA - Università degli Studi di Napoli Federico II Name, country and brief profile of the customer [up to 5 lines] Farzati Spa is an innovative SME registered in the special register of Innovative SMEs at the Chamber of Commerce of Salerno – ITALY. Farzati Spa is committed to improving the quality of life for individuals and the environment through innovative solutions for the traceability and safety of products, processes, and supply chains. Problems that the customer needs to solve [up to 5 lines] BluDev® is an advanced technology developed by Farzati S.p.A., aiming to revolutionize the traceability of "Made in Italy" products through the creation of a "Bio Digital Fingerprint" – The company wants to expand/further develop the use of this technology for the certification of smart irrigation sisytems. Brief description of the service(s) (including any customisations) [up to 5 lines] This service represents the first attempt to certify applications of smart irrigation systems to a tomato crop and strawberry. This technology utilizes principles of blockchain, sensors integration and artificial intelligence to ensure a sustainable water use in agricultural production. Service execution state Not started yet Ongoing Completed x (please check the appropriate box) Service onboarding and startup: shareable insights and experiences (if any) [up to 10 lines] We have done a first year testing of the technology in indoor and field cultivation systems. We are collecting data and fine-tuning a few parameters related to optimal water distribution, product yield and quality, and technology of water certification consumption. Service execution: shareable insights and experiences (if any) [up to 10 lines] Alligning different stakeholders to improve and test the technology in a real agricultural context is challenging. We need a second year testing/validation. Post-service analysis: shareable insights and experiences (if any) [up to 10 lines] --- 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 25 Code(s) and name(s) of the digital service(s) S00022 - Virtual training or validation of crop and weed detection Partner providing the service(s) RISE - Research Institutes of Sweden Name, country and brief profile of the customer [up to 5 lines] CrossControl AB, Sweden supports customers in making industrial vehicles and machines smarter, safer and more productive. Through operational excellence and engineering expertise, the company is a trusted partner for OEMs and system integrators across the world, providing powerful platforms for machine intelligence, communication and human-machine interaction. Problems that the customer needs to solve [up to 5 lines] The customer has developed AI-based image classification models, trained to recognize various weeds and/or crops in field images and would like to assess their generalisation properties and identify avenues for further development. Brief description of the service(s) (including any customisations) [up to 5 lines] Evaluation of CrossControl’s AI-based image classification models for crop and weed detection on RISE open Weeds database. Recommendations of methodology for further improving CrossControl’s AI-based image classification and/or object detection systems, including hardware performance review. Service execution state Not started yet X Ongoing Completed (please check the appropriate box) Service onboarding and startup: shareable insights and experiences (if any) [up to 10 lines] Contract negotiations are ongoing and there are still outstanding differences between RISE and CrossControl that must be addressed. The issues are mostly of a technical and operational nature and both teams expect that they can be bridged. Of particular importance is the fact that CrossControl, as a company focusing on high TRL levels, must feel comfortable with the setup deployed by AgrifoodTEF. Service execution: shareable insights and experiences (if any) [up to 10 lines] --- Post-service analysis: shareable insights and experiences (if any) [up to 10 lines] --- 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 32 Code(s) and name(s) of the digital service(s) S00200 AI Integration – Model Improvement / Adaption Partner providing the service(s) L-PIT - Łukasiewicz Research Network Name, country and brief profile of the customer [up to 5 lines] QZ-Solutions - technological company employing AI methods for remote soil quality and composition analysis, based on hyperspectral imaging data. Problems that the customer needs to solve [up to 5 lines] Improved classification and regression models, generalizing to datasets collected at different geographical locations and in different conditions Brief description of the service(s) (including any customisations) [up to 5 lines] Training of a range of diverse classification and regression models for different soil elements, including ensemble methods as well as deep learning models, on different splits of the datasets, and systematic assessment of their performance, taking into account and analyzing the impact of different parameter Service execution state Not started yet x Ongoing Completed (please check the appropriate box) Service onboarding and startup: shareable insights and experiences (if any) [up to 10 lines] 1) fine cooperation, including agreement for multiple digital as well as physical services for the company. 2) Specific use case, need expertise knowledge Service execution: shareable insights and experiences (if any) [up to 10 lines] Customer provided samples of their own datasets, both hyperspectral images and ground-truth data, and information about their pipeline of soil assessment based on the hyperspectral imaging, and we’ve been working closely with the company to validate and improve their approach. Post-service analysis: shareable insights and experiences (if any) [up to 10 lines] --- 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 33 Code(s) and name(s) of the digital service(s) S00054 Datasets – hybrid training Partner providing the service(s) Ł-PIT - Łukasiewicz Research Network Name, country and brief profile of the customer [up to 5 lines] QZ Solutions (Poland), technological company employing AI methods for remote soil quality and composition analysis, based on hyperspectral imaging data. Problems that the customer needs to solve [up to 5 lines] Overcome and compensate for the scarcity of annotated ground-truth data based on samples collected at different geographical regions and in different conditions with corresponding satellite hyperspectral imaging data. Brief description of the service(s) (including any customisations) [up to 5 lines] Application and testing of different data preprocessing and augmentation methods, and assessment of their impact on the performance of the classification and regression models. Service execution state Not started yet Ongoing x Completed (please check the appropriate box) Service onboarding and startup: shareable insights and experiences (if any) [up to 10 lines] It was a smooth process, including agreement for multiple digital as well as physical services for the company. Service execution: shareable insights and experiences (if any) [up to 10 lines] On the company’s request we performed a literature review of the specific state-of-the-art methods for hyperspectral data preprocessing and augmentation and their use for soil content classification, and employed some of them to prepare the data for classification models training, testing the influence of different parameters on the models performance. The work has been done in a close ongoing contact with the company. Post-service analysis: shareable insights and experiences (if any) [up to 10 lines] --- 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 34 5. Plans for integration of digital services with AgrifoodTEF’s upcoming digital infrastructure Building a TEF Digital Network A sequence of planned and executed actions is shaping the building blocks of a sustainable TEF digital network. These actions focus on preparation, validation, publication, and provisioning of digital services. These activities put a spotlight on the needs of AgrifoodTEF’s partners in reference to used digital infrastructure and are highlighting their achievements, especially in using resources effectively to reach a level of service provisioning. The common denominator of the analysed digital services (that are also in publication process and therefore assumed as ready for delivery), is focus on enabling AI-driven innovation applied directly in developed systems. Excluding broader contextual considerations, three key challenges for digital infrastructure emerge from service publications, i.e.: 1) Robust data management 2) Diverse solutions testing capabilities 3) Improving Tools and Frameworks Service publication confirms that as well are designed to utilize digital technologies and infrastructure to support the evaluation of data-driven applications, AI models, and related systems. Services share several objectives and features that aim to be also a contribution the TEF ecosystem. Catching the common Based on the above conclusions, challenges that must be addressed to ensure a cohesive and efficient digital network are linked with specific actions: 1) Robust data management Effective data management requires standardizing data pipelines. This process is being built to support services like "Data and Media Labeling for AI Training" and "Curation and Provision of Annotated Datasets". Another requirement is ensuring data security and sovereignty, that supports and enhance services like "Data Valorization and Dataspace Integration Assessment". The third requirement regarding data management is a quality assessment process that should facilitates services like "Data Validation and Processing Services.” 2) Diverse solutions testing capabilities Creating flexible testing environments should be focused on user support and enhancing flexibility of customization; this requirement is being utilized by services like "Experimentation with Synthetic Data Generation and Data Augmentation Techniques" and "Curation and Provision of Annotated Datasets". Direct AI model optimization is addressed in services such as "AI Performance Assessment and Optimization on Edge Devices." 3) Improving Tools and Frameworks Optimizing tools and frameworks involves virtualization services and services where software being tested runs on infrastructure provided by TEF. This requirement is demanded from services like "Testing of Plant Detection Models". On the lower level of infrastructure direct usage are service that required tools and solutions alignment supports services like "Provision of Cloud-Based Data Storage." 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 35 Moving forward To create a strong, sustainable TEF infrastructure network, the actions listed above need to be solved systematically. This process will operate in a loop - the infrastructure should instantly support the provision of the service itself, but with potential for further networking. The next year of work should focus on catching “low hanging fruits” already enabled by accessible digital services which could be open interfaces and already produced or being producing data products. In the coming period TEF participants should agree on taking the minimum effort that each digital service in subsequent publication should include at least one cross-linking element. The coordinating team should prepare a recommendation on which of these enablers should be indicated as a top priority and orchestrate their assembly into an ecosystem. Services that already use federating solutions, defined as above, should be promoted on front office and among partners to be seamlessly adopted horizontally by other nodes in the network. The most promising context in long-term perspective is a common data flows, regulated by data space and adopted standard, however TEF should consider solutions enabling discovery, visibility and remote access. 6. Plans for integration of digital services with AgrifoodTEF’s upcoming dataspace For the integration of AgrifoodTEF digital services with the Dataspace, there are three main levels to consider: 1. Data Discovery and Access: This involves organizing metadata storage, cataloging, and browsing capabilities for the dataspace marketplace. The primary goal here is to provide a clear framework for publishing properly structured data, ensuring it is accessible and well-organized within the marketplace. 2. Technical Integration through APIs: Once data is published, users need a consistent way to access it via enabled APIs, which requires standardized data structures and parsing methods. This means each data structure and algorithm structure should be clearly defined so that users can efficiently extract and understand the information they’re accessing. 3. Monetization and Value Exchange: This level addresses the business aspects, specifically data monetization. For instance, in the case of Pontus X (one of the dataspace technology frameworks adopted by AgrifoodTEF), built-in tracking and a connected digital wallet allow transactions between data providers and users, streamlining the process of data acquisition and payment. The following figure recapitulates the three levels described above and their interactions with the TEF. 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 36 Figure 2: Dataspace service integration. AgrifoodTEF is not a legal entity at this moment, so its dataspace is implemented as a distributed collection of datasets where each partner is individually responsible for data management and compliance. Therefore, each contributor to the dataspace must have their data, access, and monetization tools also implemented, according to common AgrifoodTEF guidelines. The role of the TEF is therefore to prepare the dataspace platform and establish guidelines, making it easier for external users to integrate and access the data reliably. In the future, we aim to create a unified structure with standardized data policies, as part of our data exploitation strategy. This will facilitate easier integration and establish a common approach for contributors, making data accessible through a shared set of standards and policies. Currently, discovery of metadata and catalog (level 1 of Figure 2) is provided by AgrifoodTEF, utilizing the ecosystem of Pontus-X dataspace platform (level 3). Considering long-term compatibility and scalability of integrating digital services with TEF dataspace, we need to implement common standards and data structures and common APIs (level 2). This will make users and developers find it simpler to access multiple datasets and algorithms without needing redundant integration efforts. This consistency would make the dataspace more attractive and usable for a wider audience. Ultimately, the goal is to reach a point where any data consumer, whether an outside organization or a project partner, can easily understand and use data across platforms through a standardized approach. Guidelines, common data structures, and shared ontologies will be essential for fostering interoperability and maximizing the utility of the dataspace. Once we achieve this, integration and monetization will be straightforward and beneficial to all stakeholders. 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 37 7. Conclusions This Deliverable outlines a transitional period in the life of AgrifoodTEF, and of its digital services in particular. There are multiple aspects to this transition. A first aspect of the transition concerns the evolution of AgrifoodTEF’s service offering. Section 3 provides an analysis of how AgrifoodTEF’s offering evolved during 2024 with respect to the situation at the end of 2023 (the latter recalled by Section 2). Highlights of such analysis concern a clear evolution in the catalogue’s contents, particularly for digital services, and the transformation of the catalogue from a mere list to an interactive selection tool integrated with the project’s Web Portal, designed to support easy exploration by prospective customers. Another key aspect of the transition which took place in Year 2 is the shift of the project’s activities from an initial phase where problems were examined, approaches were discussed, decisions were taken and –as a consequence– tools and methodologies were designed and set up, to a second phase where the tools and methodologies started to be exploited by project partners to provide services to customers. Executing services and gaining experience in doing so is the most effective way to identify ways to improve the infrastructure and processes of AgrifoodTEF: indeed, the experience gained by service providers in 2024 is already being exploited to this aim. Section 4 documents this, focusing on digital services, through examples of experiences and insights resulting from the digital services activated in 2024. A final aspect of AgrifoodTEF’s transition of Year 2 concerns its digital infrastructure, i.e. the technical backbone enabling the delivery of digital services not only during the five years of the project but also beyond them. Differently from what could be said at the end of 2023, the key elements of such infrastructure have now been outlined quite clearly, and a roadmap is available. The effort is currently being moved towards implementing what has been defined and dealing with the technical issues associated to the real-world implementation of sophisticated digital systems. Sections 5 and 6 provide an outline of these ongoing efforts. All in all, this Deliverable provides a (digital services-oriented) snapshot of a project that, having decided where it will go and what route to take to arrive there, is now starting to move in the chosen direction(s). We expect the next version of this document (due at the end of 2025) to show AgrifoodTEF moving at a brisker pace and reaching much farther. 101100622/agrifoodTEF AgrifoodTEF Deliverable D2.15 - Overall Digital Service Catalogue [M24 release] December 31st, 2024 38