A course for bee analytics and business with AI

The European honey bee is the most important insect pollinator of crops worldwide. Bees are vital to the welfare and food production around the whole world. In the EU, around 84 % of crop species and 78 % of wildflower species depend on animal pollination. Over recent years, a decline has been seen in the occurrence and diversity of all kinds of European wild insect pollinators. The decrease in pollinators would lead to the decline or even disappearance of many plant species, with significant ecological, social, and economic consequences. To tackle this threat, the European Commission’s first response was the EU pollinators initiative in 2018, which was revised in 2023 with the goal of reversing the decline of pollinators by 2030 under the strategy of the European Green Deal. (BeeAware, n.d.; European Commission, 2023)

Dangers threaten the bees, so it is important to find technical solutions to monitor their hives. In the course Short Advanced Program (SAP) called Bee analytics & business of RUN-EU European university, three universities, Burgos University (UBU, Spain), Häme University of Applied Sciences (HAMK, Finland) and Howest University of Applied Sciences (Belgium), joined forces. The core idea is to teach students analytics with artificial intelligence regarding bees and bee hives. Based on the technical solution students need to develop a business solution. The course is 3 ECTS credits taking nominally 84 hours of student time. The pedagogical model used is Design-Based Education. (Coelen et al., 2021)

Course conduct with DBE

The pedagogical model used was Design-Based Education (DBE). (NHL Stenden, n.d.) It is a form of Project-Based Learning (PBL) where the students learn the needed subjects while designing and developing an artefact or a system for a real-life problem that is open-ended.

A PBL approach requires a workplace-like environment, i.e., one that supports coworkers in setting goals, pursuing excellence, and overcoming obstacles to ensure success (Pennefather, 2022). Students need to be guided to learn in a variety of different ways. First of all, by reading other people’s works and watching videos provided by their instructors, they can discuss and debate with their peers and their teachers, both in person and remotely. Secondly, by reflecting on what they learn and presenting to others what they know, students can come up with new ideas and proposals. Finally, by developing a prototype or minimum viable product (MVP), they provide a potentially useful solution for a wide range of customers. As PBL involves diverse ways of learning, mentors play different roles along the way: from content designers and teachers to group facilitators and advisors. (Jaganathan et al., 2024)

In our case, the students first had a mandatory part in analytics and then they were given free hands to design and develop a system. The main concepts of DBE are multidisciplinary, international and intercultural collaboration, Design Thinking, sustainable education and personal leadership (Stump, 2021) These were accomplished in the SAP.

The course started at the beginning of February 2024 in Microsoft Teams. The first five weeks were online. The course started with kick-off sessions where the students got to know each other. The main teaching materials were videos giving student flexibility to learn in their own time. The sixth week was an intensive week at Burgos University where the students and facilitators joined.

In our international multidisciplinary course, we made sure to create teams that were a mix of IT and business students, along with others from different academic backgrounds. This diverse blend allowed for some great cross-disciplinary learning opportunities. With students from 21 countries and five universities, our teams were truly global. Each team had around four to six members and was mentored by a teacher from one of the organizing universities. In total, we had eight teams, making up a total of 42 students. It was a collaborative and enriching experience for all involved.

In the field of Artificial Intelligence (AI) education, Howest’s Multimedia and Creative Technologies (MCT) provided all the technical content. The course benefitted from the close collaboration of Howest MCT with Appalachian State University and their renowned AppMAIS (Appalachian Multi Agent Intelligent System) project (Stump, 2021). AppMAIS is dedicated to advancing beekeeping practices through innovative remote monitoring solutions. Specifically, AppMAIS offers beekeepers and researchers the ability to remotely monitor the growth and health of their hives.

Utilizing video banks captured directly from the entrances of beehives, AppMAIS provides invaluable insights into bee behaviour and hive conditions. Moreover, it incorporates data collected from hive sensors, including weight, humidity, and temperature, further enhancing the depth of analysis available to users.

Through the generosity of Appalachian State University, our course was granted access to the extensive video bank and sensor data curated by the AppMAIS project. This allowed our students to delve into real-world data and gain practical experience in analysing bee behaviour and hive dynamics, enriching their learning experience (AppMAIS, n.d.)

Video processing with AI

In the SAP course, students delved into video analysis using cutting-edge tools like Roboflow (Roboflow, n.d.) and YOLO (Ultralytics, n.d.). They utilized these systems to replicate the innovative solutions pioneered by Howest. With Roboflow, they practised their skills by analysing video frames and drawing detection boxes around the bees. Training a YOLO model enabled the students to apply real-time bee detection. Following this challenge, teams were given the freedom to explore and expand upon the project.

In addition to detection (Figure 1), students had to perform image classification (bees/drones) using a supervised classification method. They also raised the problem of detecting and classifying dangerous species in the hives (binary and/or multi-class classification). Moreover, they discussed how to model the relationship between the ratio of certain classes and other variables collected from the hive sensors, together with modelling the health of hives with available variables. To address these tasks, students were given materials to study basic statistics, time series, regression and logistic regression models, basics of deep learning and convolutional neural networks, as well as object detection.

Results and feedback

Seven teams decided to prototype a beehive monitoring and analysing system for beekeepers. One team built a detection system of flowers for beekeepers. The level of the systems ranged from user interface prototypes made with the Figma system to almost usable server installations. Some student teams even managed to analyse the sensor data so that both the short and long-term data could be shown to the beekeepers. All this even though the study time in SAP was limited. In the final presentation lecturer and a beekeeper gave comments online. (Figma, n.d.)

The business students of the teams had created a business model canvas for their solution. Also, they needed to think about economic requirements and create a brand book. For marketing some teams created web pages.

The intensive week started with all students in a single classroom. The students used a lot of the videos so the WLAN access point could not cope with the traffic. Fortunately, there was another classroom with another access point that solved the situation by doubling the bandwidth. This also reduced the noise level in the classroom making it more comfortable to work in. In the future, these could be taken into account when reserving classrooms.

In the oral feedback session, the students were happy about the course. The mix between IT and business students was successful; they were able to develop the whole task together and also learn from each other’s points of view. Comments from the beekeeper in the final presentation were important to the students. Overall, the objectives of the course were reached.

Since this was the second SAP executed with this method of joining technology and business cooperation the worst pitfalls had already been avoided in advance. During the feedback session, two main development areas were introduced to teachers: the IT students needed a video giving an overall picture of business related to the IT world, and the team building should have been done earlier.