There's a lot of hype around artificial intelligence, and it can be hard to sort the signal from the noise.

So why not talk to a real live human? Give us your info and we'll contact you right away.

Contact Us

Fish counting thru glas

Using AI and the DNNCam To Count Fish

If you’ve never heard the term “fish census,” you might not realize that various government regulations require dam operators to count fish. Basically, you can’t operate certain types of dams in the United States without providing a way for fish to cross the dam (known as a ‘fish ladder’). Fish need to cross the dam to reproduce, hence the regulation.

Additionally, if your company operates a dam with a fish ladder, you’re expected to provide a report showing how many fish have used the ladder, what type of fish they are, etc.

Fish ladder

A rough schematic of a fish ladder. Image courtesy California Dept. of Water. Click for more info.

Fortunately, this is the exact type of problem that artificial intelligence was meant to solve. This article explains how our DNNCam is being used to count fish. Check it out.

Why Would Anyone Need To Count Fish?

The Federal Energy Regulatory Commission (FERC) mandates that all power companies operating dams throughout the United States install a ladder or tunnel that allows fish to get up and over the high walls of the dam. Without this ladder, fish are unable to reproduce upstream. In theory, this mandate makes sense for the sustainability of fish breeds.

However, in practice, it is very challenging to monitor the ladder and comply with the requirement to report back to the FERC with data on the number and type of fish making their way up the ladder successfully. An individual employee with the power company typically has to watch the fish through a 3′ x 3′ glass window.

Fish counting room

Here’s “the vault”an underground concrete room adjacent the fish ladder. You look thru the window on the left, which peers into the fish ladder. You wait for a fish to swim by, and then determine it’s species and relative size. Then you write that down and wait for the next fish.

You can image how tedious and dull the job of monitoring fish can be, not to mention how inaccurate the data may be. The solution, of course, is AI.

Idaho Power Company Partners with BoulderAI

Recently, the Idaho Power Company and BoulderAI began working together to develop a computer and camera system to take the human component out of the fish ladder monitoring process. The system monitors the ladder at the Upper Melad River (a tributary of the Snake River near Twin Falls, Idaho), identifies fish breeds, and counts the number of fish moving up the ladder within a given period of time.

Fish counting software

Fish aren’t always easy to see, so we started with a big technology stack.

In the beginning, the BoulderAI system began with rudimentary parts–stacks of big computers and different cameras–which worked with accuracy but not perfection. Today, the system is only one small, highly-accurate camera that monitors the ladder and supplies the data automatically in a simple format that’s easily reported.

The camera, which is called the DNNCam, counts the fish through the glass, identifies the breed of fish with 96% accuracy. The camera can account for redundancy and adjust numbers accordingly. In other words, our system is capable of distinguishing between individual fish – not just breeds of fish.

The net result: A fish census that meets FERC’s mandate without expensive manpower.

DNNCam Is The Future Of Fish Counting

Fish counting thru glas

Our software can determine fish species, length, and direction of travel.

Currently, the Idaho Power Company is the only company employing the DNNCam, but it would be a relatively simple process to install and startup a similar system in other dams. With our existing technology and data, we can have a system up and running quickly.

Instead of paying a human to count fish, you can contact us…and your staff can go back to counting fish for fun.

Darren Odom

The founder of Boulder AI, Darren’s engineering resume includes consulting for leading-edge research companies, product design and development, custom software application development, and more. Darren holds numerous patents, and is a recognized expert in the fields of deep learning and AI hardware development.