Most fish monitoring programs are built around what’s historically been possible rather than what’s currently possible. That’s a natural way for institutions to operate. Programs get designed around available tools, funded, staffed, and embedded in reporting workflows. Upgrading them requires more than buying new equipment. It requires rethinking what the monitoring program is actually supposed to deliver.
The FishL Recognition system is the piece of technology that makes that rethink worthwhile. It’s not a marginal improvement on existing monitoring approaches. It’s a different category of monitoring tool that generates a different category of data product.
Conventional monitoring produces a sampled estimate of what passed a monitoring point. The FishL system produces a complete record of what passed. That’s not the same thing with better numbers. That’s a fundamentally different relationship between the monitoring infrastructure and the population it’s describing.
The distinction matters most when the population is doing something unexpected. A sampled estimate smooths out anomalies. A complete record captures them. When a migration run shows an unusual species composition, or a peak passage event at 3 AM that falls outside the scheduled sampling window, the sampled estimate misses it. The fish scanning equipment running continuously logs it, classifies it, and timestamps it.
What Data Does Each Fish Generate in the FishL System?
Here’s the complete output for every individual fish that passes through the imaging zone:
- Species classification from AI algorithms trained on thousands of verified images across 12 or more Pacific and Atlantic species
- Fork length measured computationally from the multi-angle image set
- Girth measurement derived from the same image set
- Adipose fin status, distinguishing wild from hatchery-origin fish without prior tagging
- Injury assessment including scale loss, fin damage, and visible lesions
- Tag detection for PIT-tagged or externally tagged fish
- Timestamp accurate to the passage event
- Water temperature and ambient flow rate at time of passage
- Full 18-image set attached to every record for post-hoc verification
Every one of those data points is generated automatically, without staff interaction, for every fish. Not a sample. Every fish.
How Does This Change What Hatchery Managers Can Do In-Season?
For hatchery programs dependent on accurate wild-to-hatchery ratios in the returning run, the FishL system delivers that breakdown in real time. Managers can see on any given day what the composition of the run is, adjust broodstock collection priorities, and make harvest or release decisions based on actual current-season data rather than the prior year’s model.
That’s the kind of operational intelligence that turns a fisheries program from a reactive management exercise into a proactive one.
What Does the Bonneville Dam Deployment Tell Us?
In a single deployment season at the Bonneville Adult Fish Facility on the Columbia River, the FishL system generated over 220,000 classified fish images from 12 species. That dataset was made available to the National Marine Fisheries Service and produced population health and run composition insights that the facility’s conventional monitoring program hadn’t been delivering.
Bonneville is not a small facility with a simple fish community. It’s one of the highest-volume, most species-diverse salmon monitoring sites in North America. Generating 220,000 verified image records there, in one season, without requiring an expansion of the monitoring crew, is the kind of outcome that demonstrates what the system actually delivers in real conditions.
Do You Wish to Find out More
The FishL Recognition product page has the full technical specifications. The fish monitoring expertise page covers where it fits in a broader monitoring program. Published Whooshh blog on how fish monitoring systems are transforming fisheries covers the industry context. Contact us to talk through what it could add to your program at Whooshh Innovations.
