My name is Vladimir Kulyukin. The BeePi story begins when I came across the August 19, 2013
issue of Time whose cover read "A World without Bees." That issue featured
Bryan Walsh's article "The Plight of the Honeybee" about possible effects of failing honeybee colonies. After reading this article I asked myself what I can do as a computer scientist and
a beekeeper to improve the health of bee colonies.
What exactly is the problem? The problem is that many honeybee colonies are failing. In the U.S.,
the number of hives that don't survive the winter months has averaged 28.7% since 2006-2007. During the 2014 California almond bloom, between 15% and 25% of beehives
experienced failures ranging from complete hive collapse to dead and deformed brood. Typical culprits of bee colony failures are Varroa mites, foulbrood, chalkbrood, and bad weather. Other factors contributing
to bee colony failures include pesticides (e.g., neonicotinoids), monoculture (increasing emphasis
on growing large quantities of the same crop), and transportation stress caused by
long hauls of large quantities of commercial beehives on semitrucks. Since 1 out of every 3
crops have the honeybee as their sole pollinator, the nutritional diversity of our food supply
is in danger.
It's not only the bees that are suffering, beekeepers are suffering, too. The average loss for beekeepers annually is 1 out of every 3 hives. The loss range varies from 25% to 50% per year. Nationally we've lost more than half of our beekeepers since the arrival of the Varroa mite, a parasite that now afflicts many beehives. The cost of equipment, bee packages, maintenance, and transportation is so high that profit margins for beekeepers are very small.
But, there is hope! I believe that in the future significant practical and scientific benefits
will come from transforming our bee yards into smart worlds. Think of a beehive as an immobile
robot that monitors the health of the bee colony inside, analyzes the data from its sensors,
and alerts the beekeeper of any deviations from the norm through the Internet of Things (IoT).
Why is this useful? It's useful, because human beekeepers cannot monitor their hives
continuously due to obvious problems with logistics and fatigue. It will also reduce the number of invasive hive inspections that disrupt the life cycle of bee colonies
and put stress on the bees. The transportation costs will likely be lower, because many
beekeepers now drive long distances to their far flung bee yards.
To make this vision of beehives as intelligent immobile robots a reality, we need
terrabytes of good quality data collected over multiple years and
at multiple locations. We need this data to develop automated diagnostic bee health
models. In other words, we need to put AI into beekeeping.
This is what the BeePi system is all about. BeePi is a multisensor electronic beehive monitor all of whose components fit in a standard Langstroth box that thousands of beekeepers use worldwide.
The monitor consists of a credit card size raspberry pi computer, a camera, a waterproof temperature sensor, a battery, a hardware clock, and a breadboard for sensor integration. There's also a USB microphone hub connected to the raspberry pi and three mikes.
BeePi is unique because it uses computer vision and video analysis to estimate forager bee traffic levels, which is the number of bees going in and out of the hive per unit of time. Another unique feature is sensor fusion, because the system combines information extracted from video, audio, and temperature to make estimates of the beehive's health. Finally, the monitor is designed to work with standard Langstroth beehives, not observation hives. All BeePi hardware components are completely off-the-shelf.
Our team's goal is to build two more BeePi units and to deploy them in local beekeepers' hives in the next beekeeping season to collect and analyze more data and share the findings with all interested parties. Of course, if we raise more money, we will be able to assemble and deploy more beehive monitors.
I am qualified to complete this project. I have a Ph.D. in Computer Science and have been doing R&D work in various areas of Computer Science for 20 years. I also know how to manage R&D teams to achieve concrete results. You can check out my LinkedIn profile  for more details.
BeePi has been a community learning project since 2014. I would like to thank my wonderful current and previous undergraduate and graduate students who have contributed and are currently contributing to this project: Myles Putnam, Sarat Kiran Adhavarapu, Sai Kiran Reka, Keval Shah, Sarbajit Murkherjee, Prakhar Amlathe, Astha Tiwari, and Logan Pedersen. Sai and Keval defended their M.S. theses on various aspects of the BeePi project. I regularly use collected data samples in the graduate and undergraduate classes I teach at Utah State University. References 3-7 below are our scientific publications on the BeePi project where interested readers can find the technical details of the video and audio processing algorithms..
We assembled and tested the first pilot version in September 2014. In 2015, we tested two units in two of my overwintering beehives for two weeks. In 2016, we tested four units for approximately two months. This season we've been running four units in four beehives since early May and have so far collected 125G of video, audio, and temperature data.
Our long-term objective is to develop replicable hardware designs and open source software tools  that other people can reproduce at minimal costs. This fundraiser is a step toward this goal.
1. My LinkedIn profile.
2. My BeePi GitHub repository.
3. V. Kulyukin, M. Putnam, and S. K. Reka. "Digitizing Buzzing Signals into A440 Piano Note Sequences and Estimating Forage Traffic Levels from Images in Solar-Powered, Electronic Beehive Monitoring." In Edtrs. S. I. Ao, Oscar Castillo, Craig Douglas, David Dagan Feng, and A. M. Korsunsky, Proceedings of International MultiConference of Engineers and Computer Scientists (IMECS 2016): International Conference on Computer Science, Vol. I, pp. 82-87, March 16-18, Kowloon, Hong Kong, IA ENG, ISBN: 978-988-19253-8-1;ISSN: 2078-0958. Best Paper Award of The 2016 IAENG International Conference on Computer Science. (pdf)
4. V. Kulyukin & S. Reka. "A Computer Vision Algorithm for Omnidirectional Bee Counting at Langstroth Beehive Entrance." Proceedings of International Conference on Image Processing, Computer Vision, and Pattern Recognition (IPCV'16), pp. 229-235, ISBN: 1-60132-442-1, CSREA Press. Las Vegas, NV, USA, July 2016. (pdf)
5. V. Kulyukin & S. Reka. "Toward Sustainable Electronic Beehive Monitoring: Algorithms for Omnidirectional Bee Counting from Images & Harmonic Analysis of Buzzing Signals." Engineering Letters, vol. 24, no. 3, pp. 317-327, Aug. 2016. (pdf)
6. V. Kulyukin. "In Situ Omnidirectional Vision-Based Bee Counting Using 1D Haar WaveletSpikes." Lecture Notes in Engineering and Computer Science: Proceedings of The International MultiConference of Engineers and Computer Scientists 2017, 15-17 March, 2017, Hong Kong, pp. 182-187. ISBN: 978-988-14047-3-2. (pdf)
7. V. Kulyukin & S. Mukherjee. "Computer Vision in Low-Power Electronic Beehive Monitoring: In Situ Vision-Based Bee Counting on Langstroth Hive Landing Pads." Graphics, Vision, and Image Processing (GVIP), vol. 17, issue 1, pp. 25 - 37, June 2017, ISSN: 1687-398X. (pdf)