The humble honey bee might be a diligent worker and loved for the honey it produces, but few know just how vital it is to our food chain. According to some estimates around a third of all plant-based food we eat is dependent on animal pollination, of which bees are one of the most important contributors. Take it beyond food, and three-fourths of the world’s flowering plants depend on animal pollination for reproduction. In fact, bees pollinate 70 of the approximately 100 crop species that 90% of humanity lives on.

But the honey bee is in trouble, with Colony Collapse Disorder - which sees entire hives collapse due to the desertion of worker bees - emerging as a major threat. Linked to factors ranging from invasive mites and parasites to pesticide poisoning and poor beekeeping practices, Colony Collapse Disorder might even, if unchecked, affect the world’s food supply someday.

Harnessing AI to protect the bee population

Even as researchers are trying to pinpoint the exact reasons for Colony Collapse Disorder, efforts have shifted towards prevention and mitigation, with bee health monitoring and symptom-based treatment now the main weapons. That’s where Artificial Intelligence and Machine Learning come in. Across the world, several teams from universities and startups alike are now harnessing the latest developments in these fields, along with developments in infrared and camera-based sensors, to help save the world’s bee population. 

Analysing audio and video recordings to track bee health

A team at the University of Montana has developed the Bee Health Guru app, which operates on the principle that the sound of bees changes during times of stress and ill-health. Bee Health Guru analyses audio recorded (via smartphones) from a hive and compares it against a large dataset of bee audio recordings to provide early warning of any issues. Not only does Bee Health Guru enable quicker analysis, but with just 30 seconds of audio needed, it’s logistically far easier than old-fashioned manual inspections of beehives.

A similar approach, albeit using visual data (from photographs) was adopted by researchers at the École Polytechnique Fédérale de Lausanne in Switzerland. Targeting the varroa, a mite that is considered one of the biggest threats to the world’s bee population, the ApiZoom research team trained a system using a dataset of images to recognise - and count - dead mites, helping beekeepers gauge the severity of any infestation. The team has worked closely with beekeepers to get feedback and fine-tune their algorithms. It is now hoped that this system could be deployed nationwide by the Swiss government to help create a national database of Varroa infestation. 

Global giants aid the cause

The world’s tech titans are also lending a helping hand. Apic.ai, a startup which is using Machine Learning to monitor bee colony health, based its hive monitoring project around Google’s TensorFlow framework. The tools they’ve employed include neural networks to identify bees, motion detection to ascertain how many bees enter and leave their hive, which helps give early warning for colonies that may not be in the best of health, and examining flying characteristics to determine pollen collection, and therefore availability in the surrounding areas. 

The World Bee Project, meanwhile has set up its Global Hive Network, a data collection effort that hopes to analyse honeybee health and its interactions with local conditions, weather, pesticide use, and other factors. The Global Hive Network uses data collated from various sources - weather reports, farm data, satellite imagery, and hive sensors, which is then analysed using Oracle’s database, analytics, and cloud computing platforms to monitor bee health and to detect any problem areas in time for successful intervention. 

Then there’s German startup We4Bee, which has partnered with Microsoft under the aegis of the latter’s AI for Earth programme, which explores ways AI can boost conservation and species protection efforts. We4Bee has also adopted a similar tack to other teams in this field - building ‘smart’ hives using sensors to collect data which can provide advance warning of any serious issues, and at the same time, help scientists understand bees’ response to environmental stimuli better. 

Commercialisation beckons

AI-based products that help safeguard bee health are not just for the research or institutional sectors. With beekeeping a major financial activity, this offers startups a fresh opportunity for commercialisation of their technology. Canadian startup Nectar is already on this path. Tagging themselves as providers of ‘precision’ beekeeping technology, Nectar has created a full-fledged platform for beekeepers, providing sensors, wireless data bridges, data processing using machine learning algorithms, and a dashboard to help provide actionable information in a clear, easy-to-understand way.

The importance of bees - whether as pollinators for our food supply, or as a business sector in itself, has led to a happy circumstance in which technology is now being used to safeguard the future of a vulnerable species. We can also hope that over time, as data collection and analysis improve, the exact causes of Colony Collapse Disorder are identified - and tackled. 

Image credit: PollyDot