AI TOOLS, SMART AGRICULTURE TECHNIQUES AND SUMMER SCHOOL COURSES DEVELOPED FOR OLIVE PRODUCTION: CASE OF THE DEEP FARM ERASMUS+ PROJECT APPLIED AT IZMIR, TURKEY

AI TOOLS, SMART AGRICULTURE TECHNIQUES AND SUMMER SCHOOL COURSES DEVELOPED FOR OLIVE PRODUCTION: CASE OF THE DEEP FARM ERASMUS+ PROJECT APPLIED AT IZMIR, TURKEY HI. Aybay1, M.E. Dogan2, R.A. Uzel3, P. Atakan3, N. Aygun3, M. Oladunjoye1 1Eastern Mediterranean University (TURKEY) 2Akdeniz University (TURKEY) 3Yasar University (TURKEY) Abstract This paper discusses smart agriculture techniques used for the part of the Deep Farm Erasmus+ project applied at İzmir, Turkey by the Turkish coordinator Yasar University, focusing on olive trees and olive products. The aim is to improve olive farming practices by enabling Artificial Intelligence (AI) supported decision-making through the analysis of real-time and historical data. Advanced AI technologies on the software side and modern equipment, like in-ground sensors and drones, are used to enhance olive production performance. Key achievements so far include the deployment of AI-driven models for olive farming. A You Only Look Once (YOLO) deep learning model-based disease detection system has been implemented for olive trees. Additionally, a Hybrid Gated Recurrent Unit (GRU) model has been developed for dynamic weather prediction. Integration of field data and historical weather records is successfully done which further enhances the system’s predictive capabilities. At the end of the olive case study which is expected by the end of December 2025, we believe the results will show that continuous and systematic observation of crops, dynamic weather prediction systems, use of modern farming tools, and proper use of AI techniques and tools for early detection of diseases will help sustainable and innovative agricultural practices in olive production. Keywords: Agriculture, Artificial Intelligence, Education. 1 INTRODUCTION In recent years, many studies have been published discussing how AI tools can be used for more optimized and sustainable agricultural production [1] - [3]. Following these studies, a group of faculty members from France, Italy and Turkey decided to propose an Erasmus+ project to develop new AI tools for Agriculture, combined with the use of modern devices like sensors and drones for improved productivity [5], [6]. Four developing countries from Caribbean (Haiti and Dominican Republic) and Africa (Ivory Coast and Madagascar) joined the project, choosing different crops for each country, such as bananas and cacao. It was decided that Turkish partners should work on developing AI tools and also develop an experimental case study considering olive trees. The overall coordinator of the Deep Farm project is the Institute of Advanced Industrial Technologies (ESTIA) of France. Each country has selected a crop for its use case, aiming at more sustainable and efficient farming practices using AI tools. The objective of the olive case study of the Deep Farm project being implemented in Turkey is to enhance the productivity of olive cultivation through the integration of Artificial Intelligence and modern farming tools with continuous observation of olive trees. Central to this case study was the development of AI-Driven Innovations, focusing on the creation of advanced machine learning models for both disease detection and dynamic weather forecasting, specifically tailored to the challenges faced in olive farming [7], [8]. The olive case study project aims at promoting sustainability in olive farming practices by enabling more informed, data-driven decision-making through the analysis of real-time and historical data [4]. These objectives collectively aim to demonstrate the transformative potential of AI in enhancing productivity, efficiency, and sustainability within the olive cultivation sector. As part of the contributions of the Turkish partners’ to the project, a summer school was organized at the campus of Yasar University in Izmir, Turkey. Lecturers and students from Turkey and the Caribbean (Haiti and Dominican Republic) and African (Ivory Coast and Madagascar) countries participating in the Proceedings of EDULEARN25 Conference June 30th - July 2nd, 2025, Palma, Mallorca, Spain ISBN: 978-84-09-74218-9 3507 Deep Farm project attended the summer school. The aim of the summer school was to equip students and lecturers with basic information about AI and machine learning tools and to introduce recent examples of successful uses of AI in agriculture. To ensure accessibility, face-to-face sessions organized during the summer school were also recorded and shared online with the students who could not attend the face-to-face sessions. The subsequent subsections provide an outline of the specific techniques, tools, and processes utilized to achieve the project objectives. 2 METHODOLOGY The methodology employed in the Deep Farm project case study for olive cultivation was designed to integrate advanced technological solutions with practical agricultural needs. This approach has several key stages, beginning with comprehensive data collection from diverse sources, including real-time field data and historical records, to form a robust dataset. The system architecture was developed to facilitate seamless data flow and processing, enabling the application of advanced machine learning models on the collected data, focusing on the creation and optimization of AI-driven tools for disease detection and weather prediction, specifically tailored for olive farming. The following diagram illustrates the main parts of the Deep Olive project. The left-hand side of the diagram indicates basic blocks of the data collection part of the system. The right-hand side illustrates the data processing part. Figure 1. Deep Olive Farm Architecture – Components and Communication 3508 Figure 2. Deep Olive Farm Architecture – Data Processing A summer school, which was mentioned above, was organized in İzmir in September 2024 for students and lecturers from the Caribbean and African participant countries of the project (Table 1). The aim was to give basic information on mathematical models and examples of AI applications in agriculture. The program included face-to-face training on the fundamentals of machine learning, examples of AI tools used in agriculture across various countries, and a discussion on the development of AI-based systems for weather prediction and disease detection. Laboratory sessions were organized every afternoon to get students familiar to the theoretical concepts discussed and examples given in morning sessions. The program spanned a duration of two weeks. During this time, students were encouraged to participate in another summer school program organized by an EU co-funded project named AGRIEU, in parallel. This project aims to increase participants' awareness of sustainability concepts in agri-food production, supply chain management, and the agri-food sector. The lectures were delivered by several academics who are experts in their respective fields over the two weeks. This initiative seeks to establish a multiplier effect between two projects, thereby enhancing participants' comprehension of sustainability concepts as well as the agri-food sector. Then, students from Yasar University who participated in the Summer School were given extra training on olive trees and started making observations on trees assigned to them at the Olive Research Institute premises located close to the university campus. With observation forms designed for the olive tree case study, students are performing regular olive tree inspections, recording weather and soil conditions with the help of sensors, and are taking close-up pictures of olive tree leaves for early disease detection. 3509 Students from the Faculty of Agricultural Sciences and Technologies and the Faculty of Engineering from Yasar University who attended the summer school mentioned above were provided with theoretical information about the olive tree's yearly life cycle at the Izmir Olive Research Institute, located near the Yasar University campus. Then, students were given guidelines and forms for olive tree observation and were assigned three olive trees each. They are performing regular olive tree inspections and are taking close-up pictures of olive tree leaves to be used in disease detection. The forms they are using are summarized below. The form is translated from Turkish to English by Deepl and post-processed by the authors for correction. The topics discussed during the summer school are given below in Table 1. Table 1. 2024 September Summer School Topics and Teaching Methodology Topics Format Basic concepts of statistics* Lecture and Laboratory Sessions Basic concepts in data analysis* Lecture and Laboratory Sessions Python Basics* Lecture and Laboratory Sessions Python Libraries* Lecture and Laboratory Sessions Basic concepts in Machine Learning and Deep Learning* Lecture and Laboratory Sessions Basic knowledge in olive farming** Farm visit and presentation by experts Field Observation** Farm visit and presentation by experts Chemistry and Microbiology Laboratory Application** Laboratory visit Agri-Food Industry in the EU*** Lecture Introduction to the EU, EU History and EU Policies*** Lecture Energy Consumption, Energy Policies in Agri-Food Industry*** Lecture Sustainability and Circularity in Agriculture and Food*** Lecture Agri-Food Supply Chain Management*** Lecture Traditional and Modern Technologies in the EU Agri-Food Industry*** Lecture The Common Agricultural Policy*** Lecture Food Law and Regulations*** Lecture Note. *DEEP FARM Summer School Topics; **Field Study/Lab visit; ***AGRIEU Summer School Topics 2.1 Olive tree observation guide This section explains a summary of the observation guidelines provided to Yasar University students who are currently conducting regular observation of olive trees at the Izmir Olive Research Institute experimentation field. 2.1.1 Pre-Observation Process Observation Objective: To examine the stages of the annual cycle of olive trees, changes in the growth process and the effect of environmental factors. Preliminary Studies: 1 Firstly, learn the locations of the olive trees allocated to you in the garden of Izmir Olive Research Institute. 2 Find out the features of the sensors installed in the garden and learn how to access the information that can be obtained from the sensors with the help of the experts working in the project. 3510 3 Learn basic information about the annual cycle of the olive tree and important dates (flowering, fruit formation, harvest, etc.) by researching. 4 Obtain an observation schedule and observation form. 2.1.2 Observation Process 1 Observation Site: Project-defined plot in the garden of Izmir Olive Research Institute. 2 Initial Observation: On the first day of your visit to the institute, find out the number and location of the trees in the plot of olive trees allocated to you. Record information about the general condition of the trees and whether the leaves look healthy or not for each tree separately. Afterwards, record the moisture level (separately for 20/40/60 cm depth) and similar data of the sensors located in the plot and for which you obtained information by research. 3 Weekly and Monthly observations: a) In accordance with the observation calendar, observe the trees according to the annual cycle of the olive tree in the current month, according to the monthly / every 15 days / weekly options and note the requested information on the observation form at the end of each observation. b) Record the moisture level (separately for 20/40/60 cm depth) and other important data of the sensors at each observation. c) Record information about the general condition of the trees, whether the leaves look healthy or not (healthy, wilted, diseased) on the observation form for each tree separately. 4 Taking Photographs a) Attach one photograph (remote shot) to your report to show the general condition of the observed tree. b) Also include 3 close-up photographs of the leaves from different parts of the tree and from different branches (at a distance of about 5-10 cm, similar to the example given to you). It should be noted here that, especially in trees where you see signs of disease, you should take photos of 3 separate leaves that you think are diseased. If you think the tree is healthy, 3 separate healthy leaf photographs should be taken. c) It is important that the leaves are not photographed in a cluster of leaves, but if necessary, closer than 5 cm. alone, separately from other leaves. 2.1.3 Organizing and Reporting Observation Results: At the end of each observation, report the results obtained separately for each tree and submit them to your experts in charge of the project. If you see signs of disease on the leaves or branches of the trees, prepare a separate warning report with the tree number and possible disease type. Submit the JPEG digital files of your photographs to your experts in charge of the project on the internet, separately for each tree, and always adding the tree number. Also, record the weather data on the observation day on the observation form (temperature, wind strength and wind direction, humidity, air pressure, sunny/cloudy/rainy). 3 RESULTS 3.1 Results from the Summer School The Summer School offered a highly multidisciplinary curriculum, covering a broad range of subjects, from EU agricultural policy, sustainability, and food law to data science, machine learning, and Python programming. The Summer School was attended by participants with different academic and professional backgrounds, including both university students and lecturers (Figures 3 and 4). This enriched the learning environment and encouraged knowledge exchange across experience levels. 3511 Figure 3: Education level of the participants Figure 4: Background of the participants A questionnaire was administered to lecturers and students participating in the Summer School, both before and after the event, to assess the impact of the lectures, field visits, and laboratory sessions. Participants were asked to self-evaluate their level of awareness on selected topics from the Summer School program. The evaluation was based on a scale with the following options: NA – Not Applicable; Fundamental Awareness (basic knowledge); Novice (limited experience); Intermediate (practical application); Advanced (applied theory); and Expert (recognized authority). For analysis, the scale was quantified on a 1–5 scale, excluding the NA option. Figure 5 illustrates the increase in awareness and skill levels among participants following the Summer School program. On average, participants reported a 66% improvement across all topics, as shown by the bar labelled "All topics." The highest increases were seen in topics such as "Python Basics" (over 80%), "Sustainability and Circularity in Agriculture and Food," and "Agri-Food Supply Chain Management." These results demonstrate the program's effectiveness in enhancing both digital competencies and thematic knowledge in agriculture and food systems. The improvements across all topics reflect a strong overall impact of the lectures, field visits, and laboratory sessions. Figure 5: Increase in awareness and skill level of participants after the Summer School program. Note. *DEEP FARM Summer School Topics; **Field Study/Lab visit; ***AGRIEU Summer School Topics 3512 3.2 Tools developed Two tools have been developed for the Deep Olive Farm project: a Neural Network based model for Olive peacock spot disease detection (web-based app and a mobile app) and a Dynamic Hybrid Weather Forecasting Model for olive tree cultivation [9] An experimental digital olive farm is developed at Izmir with data collection and data processing components as depicted in Figures 1 and 2. Currently, olive tree observations are being made by students taking part in the Deep Olive Farm project. 3.3 Results obtained by students during observations This subsection discusses some of the results obtained during ongoing olive tree observations by students taking part in the Deep Olive Farm project. As a sample, a summary of observations on one of the olive trees in the farm is provided below. Students visit the olive farm, observe the assigned trees, and complete a form for each one. They also take several close-up photos of the leaves, focusing on damaged or unhealthy ones. Once they finish their observations, they take a photograph of the tree from a distance. Tree number "B9" is selected as the sample tree. This tree is located near the entrance of the farm, making it easily observable. Figure 6. Photograph of the B9 tree from a distance. On this sample tree, based on observations made by students on April 24th at around 4:00 PM, there were no abnormalities on the trunk of the tree, but some issues were noted on the leaves. Although the majority of the leaves remained green, the number of yellow and brown (dead) leaves was rapidly increasing. Students also discovered traces of Euphyllura olivina, known as the olive psyllid, on this olive tree. It is easy to identify them by their distinctive markings found around the leaves and flower buds. 3513 Figure7. An olive tree with strong traces of olive psyllid (Euphyllura olivine) around flower buds More importantly, many of the leaves exhibited ring-like marks, suggesting the presence of a fungal disease (Spilocaea oleaginea) known as olive peacock disease. There were no fruits on the tree during the observation day, so the health of the fruit section was not applicable. Figure 8. Olive tree leaves strongly indicate possible fungal (Spilocaea oleaginea) olive peacock disease The “Filiz: Agricultural Sensor Station” was installed in the farm on August 22, 2024, and the sensor records various data from its environment every hour. Data includes air temperature, air humidity, soil surface and soil temperatures (at depths of 20 cm, 40 cm, 60 cm), soil humidity, soil moisture (20 cm, 40 cm, 60 cm), wind speed and direction, light intensity, and precipitation. Table 2 indicates data recorded from the sensor on observation day, May 5, 2025: 3514 Table 2. Data values from Filiz: Agricultural Sensor Station on observation day (24.04.2025 – 4:00 PM) Sensor Parameters Data Values Air Temperature - °C 21,9 Air Humidity - RH% 52 Soil Surface Temperature - °C 21,73 Soil Temperature - °C - 20 cm 18,6 Soil Temperature - °C - 40 cm 17,6 Soil Temperature - °C - 60 cm 16,68 Soil Surface Humidity - RH% 62 Soil Moisture - RH% - 20 cm 42,41 Soil Moisture - RH% - 40 cm 19,8 Soil Moisture - RH% - 60 cm 28,07 Wind Speed - km/h 1,35 Wind Direction W Light Intensity - W/m² 115,73 Precipitation - mm 0 Data from observation day and students’ observations show that it was a warm, slightly cloudy and windy day. 4 CONCLUSIONS Considering the educational achievements of the project, the Summer School successfully attracted students and lecturers from different countries and with different levels of experience and academic backgrounds. Managing to engage such a diverse group is a notable achievement, as it is often extremely challenging to design content that resonates across varying levels of expertise. The significant increases observed across all thematic areas in Figure 5 demonstrate the program’s effectiveness in delivering meaningful learning outcomes to a broad audience. In particular, the program played a key role in equipping Yasar University students with the foundational knowledge and practical skills necessary for conducting field observations, ensuring that they were well-prepared for hands-on learning activities following the theoretical sessions. ACKNOWLEDGEMENTS This work is supported by the EU Erasmus+ project Capacity Building for Digital Agriculture in HEIs in Third Countries (Deep Farm). The authors would like to thank Mr. Alperen Kızılkulak for his contribution to the project by giving pre-observation training to Yasar University students and by organizing and supervising observation sessions at the experimental field. REFERENCES [1] I. Attri, L. K. Awasthi, and T. P. Sharma, “Machine learning in agriculture: a review of crop management applications,” Multimedia Tools and Applications, vol. 83, no. 9, pp. 12875–12915, 2024. [Online]. Available: https://doi.org/10.1007/s11042-023-16105-2 [2] S. Mourtzinis, P. D. Esker, J. E. Specht, et al., “Advancing agricultural research using machine learning algorithms,” Scientific Reports, vol. 11, no. 1, pp. 17879, 2021. [Online]. Available: https://doi.org/10.1038/s41598-021-97380-7 [3] G. Gupta and S. K. Pal, "Applications of AI in precision agriculture," Discover Agriculture, vol. 3, p. 61, 2025. [Online]. Available: https://doi.org/10.1007/s44279-025-00220-9 3515