Beehives can be “optimized” using data and models

Bees make a significant contribution to biodiversity. Pollination is indeed an important stage in the life cycle of plants, and bees fully contribute to this. However, the threats to biodiversity today indicate the extinction of pollinating insects, in particular honey bees (e.g.Apis mellifera residence in Europe). This disappearance would be catastrophic for humans: it is estimated that almost 35% of our food depends on pollination services provided by bees!

Colony collapse syndrome disorder of disintegration of the colony or CCD, ie systematic abandonment of hives) is a phenomenon that has become repeated in colonies of honey bees. Thus, there is a significant reduction in bee colonies around the world; this can lead to the loss of up to 90% of the hives. As for honey bees, the reasons can be many: toxicological, parasitic, viral, even carnivorous, with the appearance in recent years of the Asian hornet (Vespa velyutina) then from the eastern hornet (Eastern Vespa) in the hexagon.

Beekeeping beekeepers play a key role in species conservation. To help them, they now have less invasive solutions for monitoring and predicting the health of hives. In particular, this article is devoted to two of our research papers that use IoT technologies. Internet of Things or IoT) related to artificial intelligence, computer models and modeling to help beekeepers in their business practices.

Behavior of bees at the entrance to the hive

The first key to monitoring and predicting the health of hives is to model the behavior of bees; this allows you to identify the strengths or weaknesses of the hive caused by disease, hunger or predators. In this way, we can understand what is happening inside the hive by observing what is happening outside. For this, images and videos are rich sources of information that need to be used, non-destructive, and pose a serious scientific and environmental challenge. However, trajectory modeling has not yet been developed, which prompted our work.

Our first work focused on detecting bees, as well as modeling their trajectories. The idea was to record the flight paths of bees separately, and then characterize the rhythm of general activity in front of the hive to derive observations that will be consolidated and provided to scientists for cross-reference with beekeeping data. (famine, predator, drought…).

In practice, this included recording the movement of bees with a fixed non-invasive camera, surveying the entrances and exits of insects. Using a high-resolution sensor and a high frequency of shooting, image processing methods allow you to isolate the bee from the background.

Figure 1: Body center (green dot) and offset (red, blue, purple and yellow lines) of two bees, calculated from several images
Figure 1: Body center (green dot) and displacement (red, blue, purple and yellow lines) of two bees, calculated from several images – Gregory Zakharevich and Baptist Magnier

Next, bee counters are extracted from each frame of the video, which allows you to identify the center of each bee (green dots on the bees in Figure 1). Then the orientation of each bee is represented by an ellipse (associated with the shape of the bee). The orientation and size of the ellipses in the image allow you to calculate the movement of bees between different video images (blue and red lines for bee 1 and yellow and purple lines for bee 2).

Figure 2: Graph of conflicting (left) and consecutive (right) trajectories
Figure 2: Graph of conflicting (left) and consecutive (right) trajectories – Gregory Zakharevich and Baptist Magnier

Indeed, video makes it easier to track these objects of constant size and orientation than deformed objects. This method also avoids confusion of bees and calculation of aberrant trajectories, as shown in Figure 2.

In this way, various traces are recorded. The Figure 3 is the result of a video containing 1755 images. It shows the trajectories of bees entering the hive (green lines), exits (red) or just passing in front (blue). Incorrectly defined trajectories are also shown in blue. Based on these data, it will be possible to study and classify the behavior of bees.

Figure 3: Monitoring of bee trajectories in front of the hive, based on 1755 images.  Green lines: bees enter the hive;  red lines: bees leave the hive;  blue lines: bees passing in front of the hive and poorly defined trajectories
Figure 3: Monitoring of bee trajectories in front of the hive, based on 1755 images. Green lines: bees enter the hive; red lines: bees leave the hive; blue lines: bees passing in front of the hive and poorly defined trajectories – Gregory Zakharevich and Baptist Magnier

In the future, the behavior of bees can be further interpreted by supplementing the study with machine learning data and semi-controlled AI method.

Physical characteristics of the hive

The second key to making a decision for the beekeeper is to analyze the internal condition of the hive. Here, our team uses related weights combined with data from multiple sensors (such as relative humidity and indoor temperature) to analyze changes in the weight of each hive, as well as activity videos on the take-off board (as shown above).

The BeePMN project, led by our team in partnership with USEK in Lebanon, ConnectHive in France and l’Atelier du miel in Lebanon, contributes to the physical characterization of the hive using shared apiaries from shared databases. We have proposed a methodology based on the recognition of characteristic patterns in weight data recorded by hive weights. The reason may be an increase in weight, then a plateau followed by a drop in weight, which would correspond to the departure of bees (eg, swarming).

Subsequently, the collected data were evaluated and processed using algorithms, which allowed to identify recurring patterns associated with events occurring in the hive.

Related beekeeping

Combined with other data, these two examples can eventually be integrated into a generalized hive monitoring system. Computer models (presented in Figure 4) are automatically triggered by a series of predefined notifications that invite the beekeeper to action. For example, information for the beekeeper about the need to feed the bees may be caused by the weight passing below the control value during a certain period in autumn or winter.

Specifically, since the task of beekeeping cannot be automated and human intervention is mandatory, the proposed system will simply help and guide the beekeeper to plan and more accurately perform several relevant tasks: breeding and breeding new colonies, feeding colonies, adding rivals (top of honeycomb hive ), planning sanitary operations, pest control (eg varroa), planning operations such as hibernation, etc.
In this way, the beekeeper will be able to easily control his colony, perform his routine tasks, respond to alerts in the event of possible malfunctions and forecast their future supply needs.

Figure 4: hives, sensors, AI and smartphones / tablets form a generalized monitoring system
Figure 4: Hives, sensors, artificial intelligence and smartphones / tablets form a generalized tracking system – Gregory Zakharevich and Baptist Magnier

These models are based on business rules created with the help of experts in the field. It is also anticipated that these business rules may evolve over time thanks to the contribution of the beekeeping community that uses the system.

Finally, all of the above will be organized and presented in a user-friendly interface on a smartphone or tablet, based on the principles of gamification.

This contribution will improve the experience of amateur and professional beekeepers, reduce the risks of apiary operation and open the door to other available data (detailed weather events, flowering maps, humidity, bee colony behavior, etc.) to expand opportunities. simulation. Finally, the contribution of the latest generations of digital methods, including modeling, should pave the way for beekeeping with greater precision to minimize invasive and synthetic treatments.

This analysis was written by Gregory Zakharevich and Baptist Magnier, both professors at the Institute of Mines and Telecommunications in Ales.
The original article was published on the website of Fr. Conversation.

Declaration of interests
● Gregory Zakharevich received funding from Campus France for IMT Mines Alès