AI-powered fish biomass estimation
Know your average weight, size distribution and total biomass without netting a single fish. AquaVyon measures the whole population continuously, in RAS and sea pens, across five species, so feed, stocking and harvest decisions are based on real numbers, not a stressed sample.
Why manual biomass sampling falls short
Netting a handful of fish and extrapolating to the whole tank or pen is the industry default, not because it's accurate, but because it's the only option without vision-based tools.
Small samples, big error bars
A hand-caught sample of a few dozen fish out of tens of thousands rarely reflects the real size spread of the population, especially once size variance grows mid-cycle.
Stress every time you sample
Netting and handling stresses stock, which is exactly why most farms sample as infrequently as possible, leaving long gaps where biomass is pure guesswork.
Data that's already stale
By the time a manual sample is weighed and logged, feed and stocking decisions have already been made on last week's numbers, not today's.
How AquaVyon estimates biomass
The same computer-vision pipeline runs whether footage comes from a single uploaded video, a scheduled on-site survey, or an always-on camera install, the delivery mode changes, the measurement doesn't.
Camera captures the population, not a sample
Underwater cameras (or an uploaded video, for the no-install path) observe fish passing naturally, so the model sees the whole tank or pen over time rather than a stressed, hand-caught subset.
Computer vision measures length and condition
A vision model trained on turbid RAS and pond water, low light and dense schools estimates length and body condition per fish, then converts to weight using species-specific length–weight relationships.
Aggregated into a population-level estimate
Individual measurements roll up into an average weight, a full size-distribution curve, and a total biomass figure, updated every survey or continuously, depending on your delivery mode.
Benchmarked against your own scale
Output is validated against manual weighings on your farm and species at onboarding, and re-checked as fish grow, so you can verify the number yourself rather than take it on faith.
Works across systems, RAS and sea pens
Recirculating systems and open-water pens fail in different ways, the vision models are tuned for each rather than treating water as one generic environment.
RAS
- Camera + edge unit installed in-tank or in-raceway
- Handles turbid, high-density recirculating water
- Pairs with existing SCADA/sensor data rather than replacing it
Sea Pens
- Camera rig mounted in-pen, weather- and biofouling-resistant housing
- Tuned for variable turbidity and low-visibility open water
- Works alongside welfare/behavior monitoring in the same feed
Choosing between the two systems for your operation? See the RAS vs. Sea Pens comparison.
Species coverage
Length–weight models are tuned per species, not generic across finfish. Five species are supported today; models are retrainable if you farm something else.
Tilapia
RAS · Ponds · Raceways
Kingfish
RAS · Sea Pen
Salmon
Sea Pen · RAS
Trout
RAS · Raceways · Flow-through
Sturgeon
RAS
Farming shrimp instead? See Shrimp Intelligence for post-larvae counting and biomass.
Start with a video. Scale to continuous.
You don't need a permanent camera install to get a biomass number. Upload footage for a free read in 96 hours, move to scheduled on-site surveys, or go fully continuous when you're ready. See the full ways to work with us.
Biomass estimation FAQ
How accurate is AI biomass estimation compared to manual sampling?
AquaVyon typically lands within ±3–5% of manual weighing, because every fish that passes the camera is measured rather than a small hand-caught sample. We benchmark against your own scale on each farm and species during onboarding, and re-check as fish grow.
Does biomass estimation work in turbid or low-visibility water?
Yes. The models are trained specifically for turbid RAS and pond water and low light, not only clear sea cages. This is one of the main gaps in vision-based biomass tools built for salmon-only, clear-water conditions.
What species does AquaVyon support for biomass estimation?
Tilapia, Kingfish, Salmon, Trout and Sturgeon today, across RAS, raceways, ponds and sea pens. See the species pages below for system fit and typical harvest weights.
Do I need to install cameras to get a biomass estimate?
No. You can start with a single uploaded video and get results back in 96 hours, zero hardware, zero install. Scheduled on-site surveys and always-on continuous monitoring are available as you scale up.