Why I'm building BeeSight
The story behind the project, and what I'm trying to solve.
I started keeping bees when I was eight years old. My dad and I caught a swarm in a clay pot in my hometown of Ingiriya, Sri Lanka, then transferred it to a traditional hive. I was hooked immediately: watching the workers come back loaded with pollen, listening to the hum on warm afternoons, learning how each colony had its own personality.
Over the years, I learned that beekeeping is part intuition, part experience. You develop a sense for when something's off: a change in the sound, unusual behaviour at the entrance, or just a gut feeling that things aren't quite right inside. But gut feelings only get you so far, especially when you're managing more than a couple of hives, or when problems develop silently between inspections.
When Varroa arrived in Australia, everything changed. Suddenly, the risks were higher. An infestation you miss for a few weeks could wipe out a colony. Traditional monitoring—alcohol washes, sticky boards, visual checks—felt inadequate. Too slow, too disruptive, too inconsistent.
I found myself wishing for something better: a way to keep watch without constantly opening boxes, a system that could flag problems early, before they turned catastrophic. Something practical, reliable, and built for real-world beekeeping.
That's why I started building BeeSight.
What BeeSight actually is
BeeSight is an early-stage prototype—a multimodal sensing system for beehives. It combines:
- •Thermal imaging to track brood nest temperature patterns
- •Humidity sensors to monitor internal moisture levels
- •Acoustic monitoring to pick up changes in hive sound
- •Video capture to track bee traffic and activity at the entrance
The idea is simple: catch problems before they become disasters. A temperature drop overnight. Unusual traffic patterns. Acoustic changes that signal stress or queenlessness. The kinds of things experienced beekeepers notice—but automated, so you don't have to be there watching 24/7.
Right now, I'm working on the machine learning pipeline to detect anomalies automatically. I'm also refining the web dashboard and making sure the whole system actually works reliably in a real hive (which is harder than it sounds).
Where things are at (December 2025)
Prototype v0.3 is assembled and running. Thermal camera captures frames every 10 minutes. Humidity and temperature sensors are integrated. Video pipeline is working. Web dashboard shows live data and basic charts.
The over-the-air update system is functional in internal tests, which means I can push firmware updates without physically accessing the device. That'll be critical once units are deployed in the field.
I'm currently refining acoustic monitoring (there's a lot of noise to filter out) and building the ML pipeline for thermal anomaly detection. Battery life is looking decent, targeting 2-3 months between charges, but that depends heavily on how often the camera fires.
Early field test is planned for January with three hives. That'll be the real test of whether this thing holds up in actual conditions: heat, humidity, bees building comb around it, etc.
The problems I'm trying to solve
Manual checks are disruptive and slow
Every time you open a hive, you're interrupting the colony. Temperature drops, bees get defensive, routines are disrupted. With BeeSight, you can check in remotely without disturbing them.
Problems develop between inspections
A lot can happen in a week. Queen failure, temperature swings, early signs of disease—if you're only checking every 7-10 days, you might be too late. Continuous monitoring catches these earlier.
Varroa detection is inconsistent
Alcohol washes and sticky boards work, but they're labour-intensive and give you a snapshot at one moment in time. I'm gathering thermal and video data to eventually train models that can spot mite patterns automatically.
Scaling apiary management is hard
If you're managing 50+ hives, manual monitoring doesn't scale. You need systems that give you an overview and flag what needs attention, so you can prioritise where to spend your time.
Looking for collaborators and testers
I'm building this in Adelaide, South Australia, but I'd love to work with beekeepers and researchers elsewhere. If you're interested in testing an early unit, have ideas for features, or just want to chat about bees and sensors, reach out.
Particularly keen to connect with:
- •Commercial beekeepers who could benefit from multi-hive monitoring
- •Researchers working on Varroa detection or hive health modelling
- •Anyone willing to help test early prototypes and provide honest feedback
BeeSight Built in Adelaide, South Australia
[email protected]