Project: Smart beehives and camera traps for bees and wasps (BE-HIVE)

Description: A beekeeper has an interest in the healthiest bee population possible. To achieve this, traditionally the hive must be inspected regularly by the beekeeper. This causes stress for the bees. Other stressors include diseases, parasites and pesticides. Environmental factors affecting the health of bee population are weather conditions, microclimate, plants in the environment and flowering time. The condition of a bee population can be understood from their behavior and activity. By correlating their behavior with other known information (buildings, plants/pollen, weather) we can infer the most favorable conditions for the placement of hives, and to make sure that bees operate in a healthy environment. Having bees’ populations healthy is particularly important nowadays, due to climatic change which is threatening life on earth as we know it, via the possible biodiversity collapse.

In this project, we employ state-of-the-art sensing technologies (noise, temperature, humidity, optical and thermal cameras, camera traps), together with state-of-the-art computer vision technologies (deep learning, DL) and remote sensing (aerial photography from drones) to create smart beehives, monitoring bees’ behavior and population numbers in real-time, examining in real-time potential threats to their colonies (anomaly detection). This allows to understand their well-being, react fast in dangers, especially in relation to climate change, making the life of beekeepers easier. Remote sensing allows to map the nearby environment (e.g. up to a distance of 5 kilometers) in terms of plants/flowers while DL allows to monitor the movement of bees inside and in/out of the bee hive, observing which type of pollen (and correspondingly, flowers) they carry/visit. Finally, camera traps will be strategically placed near significant sources of nearby flowers, to understand where and when bees travel to locate food, as well as whether this creates conflicts with nearby (wild-)bees located in the larger area under study.

The high-level objectives of the project are the following:

  1. Understand the optimal environmental conditions, inside and outside the hive, for bees to be healthy and function properly.
  2. Assess the impact of possible stressors which affect the bees negatively.
  3. Examine the relationship between food availability and bees’ health and productivity in terms of honey production.
  4. Understand the competition among honeybees of various colonies or local wild bees.
  5. Understand the impact of climate change on bees’ populations in Cyprus.
  6. Detect anomaly behavior of the bees, which indicates threats and dangers to them.
  7. Understand the best strategic placement of beehives, to protect the existing ecosystems and also to ensure bees’ health and well-being.
  8. Understand the whether bee colonies perform better in urban or rural environments.

From a research perspective, the objectives listed above can be translated into the following research questions:
  1. Can commercial beehives become fully automatic, fine-tuned for optimal health of the bee colony?
  2. Which sensors shall beehives include and how should they be installed for maximum accuracy?
  3. Can state-of-art DL models be used for counting large numbers of bees in real-time from single images?
  4. Can state-of-art DL models be used for estimating bees’ populations by considering multiple consecutive frames (i.e. video)? This means that objects from one frame need to be recognized in subsequent frames and not be counted twice
  5. Can we use smart sensing technology to model in space and time the bees’ movement and transition when foraging?
  6. Can we use technology to detect anomaly behavior by the bees?
  7. What species of flowers are visited by what species of bees at what time?
  8. Can we detect swarming of bees, if and when it occurs, by using cameras?

Collaboration with: CYENS LEAR MRG, CYENS MakerSpace, EMME-CARE Center

Techniques used: Internet of Things, Computer Vision, Deep learning, GIS and geospatial analysis, Aerial photography, Machine Learning, Open hardware.

Started: March, 2022

Status: On-going