The Larcos PL-Counter
On March 2, 2012, at Aquaculture America 2012 (Las Vegas, Nevada, USA), I interviewed Will Flanery and Eric Steimle (pronounced, “sty-mil”), co-founders of Larcos Aquaculture, which displayed a prototype of its Larcos PL-Counter at the trade show.
Shrimp News: Give me a quick description of the Larcos PL-Counter.
Will Flanery: It’s a real-time, flow-through, video-based counting system that uses computer imaging recognition to count postlarvae. With our current optics, it can be trained to recognize any object from about 3 to 200 milliliters. It sees, photographs and counts postlarvae as small as PL-5s, even differentiates the live ones from the dead ones.
Eric Steimle: Our counter does away with nets, water chilling and statistical guesswork. The postlarvae never leave the water. Accuracy consistently exceeds 95%, and we’re usually closer to 99%. The technology evolved from image-recognition technology that we developed for zooplankton research. It’s lightweight, portable, battery-operable and available in several configurations to meet your specific needs.
Shrimp News: People have been trying to do this for a long time. What makes your system different?
Eric Steimle: Basically, they were missing two things: Will and Kurt (our Senior Software Engineer). Other people try to take technology for counting fish and scale it down to count tiny, mostly-transparent PLs, and their techniques don’t work at that scale. We’re used to imaging zooplankton 600 feet underwater, so for us counting seven-millimeter-long PLs on land is easy.
Will Flanery: Eric’s right; we have a great team, and Kurt is literally one of the best in the world at writing this kind of image processing software. We also benefit from the speed of modern cameras and computers, but in the end, we just know some tricks that our competitors don’t.
Shrimp News: Who’s the market for the counter?
Eric Steimle: Shrimp hatcheries and shrimp farms! Hatcheries need to count PLs when they are harvested and when they are shipped to shrimp farms. Farms need to count PLs to make sure they got the correct amount from the hatchery. In the past, there has always been a little contention between the hatchery and the farm over the number of PLs in a shipment. Our counter ends that contention. Hatcheries no longer have to give farms 10% more than what’s on the order, just to make sure the farmer gets what he ordered. Farmers can also use the counter when they stock their ponds to make sure they don’t understock or overstock.
Shrimp News: Shrimp hatcheries use large quantities of PLs. Let’s say a hatchery had an order for three million PLs. How long would it take your counter to count them?
Will Flanery: In our trial run at Shrimp Improvement Systems (SIS), a large shrimp hatchery in the United States, we counted about 50,000 PL-10s a minute. If you were counting three million PLS, it would take about an hour, with no stress to the animals. It’s automated; once you set it up, Larcos will be able to close a valve when your order is filled, so you could go to lunch and come back when it’s finished. No labor, no arguments.
Shrimp News: How much does the Larcos-PL counter cost?
Will Flanery: We don’t know what the final price will be for the production version, but it will be cost-effective enough that farms won’t be able to afford to NOT have one. This year, we’re just looking for a few beta testers to help us fine-tune things like the user-interface, size, shape, mounting options and usage practices. For $20,000, beta-testers will get a functional prototype with full technical support and software. We will come to your facility, set it up and make sure you know how to use it. When the production version is ready, we’ll swap that for the prototype. We’ll use all the information that we gather during the beta tests to develop the final production model, which we hope to have ready by the spring of 2013.
Shrimp News: What will be the biggest changes from the prototype to the production version?
Will Flanery: The production model will not require a computer. All the processing will be done within the equipment. It will be tougher, faster and use less power. We’ll also add features like the ability control valves and run off batteries.
Shrimp News: Will the user still be able to feed the data from the counter into a computer?
Will Flanery: Yes, we’re currently working with people to see how they use the counter in the hatchery and on the farm. We’re thinking about putting a Wi-Fi module into the counter, so that a technician with an iPad or Toughbook (Panasonic’s rugged tablet) could monitor the progress of the count and even look at images of the PLs as they passed through the counter, which runs on our proprietary software. Camera technology has gotten to the point where it can handle these high data rates. When we get rid of the computer (more below), we’ll be able to count twice as fast.
Shrimp News: When you get rid of the computer, how will the users get the data into their computers?
Will Flanery: You could use a thumb drive to draw the data off the counter or plug a computer directly into the equipment. Some of the people that we’ve talked with at this show don’t want to have to connect the counter to an expensive computer. There’s lots of water in hatcheries and slippery spots on farms. A lot can go wrong. A military-grade, waterproof, USB connector will probably work best.
Eric Steimle: Another advantage of the system is that you can use the PL count as a digital receipt. If a hatchery is accustomed to giving its customers ten percent more than they order, with the Larcos PL-Counter, you could just give them a couple of percent over the order to account for mortalities during transport. Hatcheries will be able to deliver the PLs with a digital receipt that shows a picture of everything that went through the counter. On an order of 100,000 PLs, the hatchery would know if there were 2,000 dead animals and could send 107,000 animals to account for the mortalities. A farm receiving that order of 100,000 PLs could run them through its Larcos Counter to check the count. If it found there were 10,000 thousand dead animals, it would only have to pay for the 95,000 live ones. Hatcheries and farms could pass their digital receipts (counts) back and forth. Everybody would be happy, trust would be built and hatcheries would have less difficulty collecting their money.
Shrimp News: What are some of the other advantages of the system?
Will Flanery: There’s no handling, no stress, no netting. You set it up with a syphon pump or a valve on the side of a harvest tank and just let it run.
Eric Steimle: You’re moving your animals anyway; you might as well be counting them in the process. The next feature that we plan to add into the software is a sizing system, so that instead of just counting the animals, we would be able to provide a histogram of the sizes that would show, for example, the number of PL-8s and PL-12s in an order for PL-10s.
Will Flanery: The system reduces mortality and labor, and the accuracy of the count helps you feed properly and predict your harvest. Proper feeding also improves water quality and lowers costs.
Shrimp News: How do you plan to prevent other people from copying your equipment?
Will Flanery: That’s why we’re being careful with this prototype. That’s why we’re only doing beta tests with people we trust. We’re going to incorporate an FPGA (Field Programmable Gate Array) chip in our production model. It’s a chip that’s extremely programmable and can do parallel processing. They’re used in extremely high-speed data processing, like image recognition. With an FPGA chip, the data is locked in—military grade—you can’t crack it. You could buy all the equipment that we use, you could see all the data, but you would not be able to make it work. You might even be able to reproduce the circuit board and chip, but you would not be able to make a clone of our system because there’s no way to suck our software off the chip.
Shrimp News: When will the FPGA chip system be ready?
Will Flanery: Hopefully, by the end of the year. We already have the software running on a PC. Now it’s just a matter of getting our FPGA expert to transfer the software to our chips.
Shrimp News: Will that raise the cost of the system?
Will Flanery: Actually, it will lower the cost. High-speed, industrial, cameras are quite expensive. Our custom circuit boards will cost less because we’re going to put a light-sensing chip right on the circuit board. The big advantages, however, are increased reliability (by eliminating cables and connectors), durability, and speed.
Shrimp News: What is in the current Larcos PL-Counter that is taking the pictures?
Will Flanery: An industrial high-speed, line-scan camera. It’s what they would use in a factory to inspect, for example, Coke bottles flashing by on a high-speed production line. It takes up to 36,000 pictures a second.
Eric Steimle: One of the things that we’ve been doing at this show is looking for investors, so that in the next year, we can incorporate the FPGA chip into the system.
Shrimp News: Where did the name Larcos come from?
Eric Steimle: Larval counting system.
Shrimp News: When did you have that “eureka” moment, when you realized that all your background and experience could best be utilized in a shrimp larvae-counting system?
Eric Steimle: From my background in oceanography and doing some robotic and sonar work, I had a Didson, high-end, 2-D sonar. It produces a video-quality image and can see animals in water. I was asked if it could do biomass calculations of aquaculture ponds, so I went up to Auburn University and started to learn about fish density in catfish ponds, but it was very difficult to get accurate readings on the biomass in the ponds because the animals are constantly moving and you just can’t see the whole pond at once. I also went down to Cartagena, Colombia, to Oceanus, a shrimp farm, and worked with those guys. They had forty-acre ponds, holding fifty tons of shrimp. We got beautiful images, we could see the shrimp when they accumulated in patches, and we discovered a lot of things that helped Oceanus manage its ponds, but it was too difficult to do biomass calculations from the images. When we were there, however, we learned a little bit about shrimp farming. When they were stocking ponds, we saw how much of a pain it was to count the PLs. I knew that I could put together a system that could count PLs. Will agreed, it wouldn’t be too hard. That was three years ago, but we just sat on the idea for a while. Then in 2011, I thought “holy cow”, let’s do it.
Will Flanery: It really started ten years ago when I worked on a high-speed zooplankton imaging system for in-situ (submerged)plankton research in the Gulf of Mexico. I reduced the system down to only 10% of its original the size and weight, and doubled the image quality and speed. For years I pushed to get the technology commercialized, but that never happened. Back in 2007, went to work for a big R&D company because they promised to commercialize our work from the university. After several years, there was still no progress in that direction. As funding from federal and state governments dried up, and military funding was increasingly competitive, I suggested pursuing funding for research on equipment for aquaculture, a billion dollar industry. My boss agreed to fund me to go to Aquaculture America in San Diego in 2010, but three days before the conference, some one higher up changed his mind. I had already bought a plane ticket and arranged lodging with a friend, so I called in sick for a week and went to San Diego on my own dime. I didn’t have enough money to go to the talks, so I spent three days walking around the tradeshow talking to everyone. Eric was there, too, and shrimp guys were there, and that’s where we hatched our plan.
Eric Steimle: I was at the San Diego meeting, sharing a booth with Sound Metrics, trying to determine how we could use our expertise for aquaculture, but Sound Metrics’ equipment (the Didson sonor), even though it works very well, was too expensive for aquaculturists. Then last year, we shared a booth with Zapco, which makes oyster cages and related equipment, in New Orleans, and this year we’re here on our own.
Will Flanery: Six months after the San Diego meeting, I drummed up the courage to go out on my own as an electronics consultant. I had enough work to pay the rent, and on the side worked on the PL-counter. Six months after that we were at the WAS meeting in New Orleans (2011) with our first prototype. It would count, but could not tell the difference between a fish and a shrimp.
Eric Steimle: We made a lot of contacts with shrimp farmers in New Orleans, all of them enthusiastic about a PL-counter. Since then, we’ve done bunches of trials with our counter, changed the software and matured the product.
Shrimp News: Can the PL-counter count juveniles, which may weigh less than a gram, and market-size shrimp, which weight over 10 grams?
Eric Steimle: Sure we can count anything small enough to flow through our system. We’ve done PL-10s; we’ve done juveniles that were point three grams; and we’ve done five-gram, seven-gram, ten-gram and fifteen-gram shrimp. It just gets easier the bigger they get. It process from 24 to 240 gallons of water and shrimp a minute. We’ll be able to configure systems to meet the specific needs of hatcheries and farms. It can even do lobster and crab larvae.
Shrimp News: Can you feed the data from the PL-counter into a database or a spreadsheet program for further analysis?
Will Flanery: Sure, you can save the data file, so you have a complete image of everything that went through the pipe. It can be saved and reprocessed anyway you want.
Shrimp News: Does the counter produce images or just numerical data?
Will Flanery: It produces pictures—black and white pictures—with our own image compression software. It also sorts the images and gives your counts and other data.
Shrimp News: How many beta testers are you going to use?
Will Flanery: I think we’ll only be able to handle four or five. To begin with, we’re looking for hatcheries that work with multiple farms.
Background on Will Flanery: I’m an electronics engineer specializing in embedded systems. That’s any electronics product that has a brain, usually a computer chip that’s doing some processing. Your microwave oven has an embedded chip that does the timing and makes sure the door is closed. Anything that’s smarter than a light switch and dumber than a PC is an embedded system. I have a lot of experience with optics. I made holograms in my parents’ garage when I was in high school and early college. When I was at the University of South Florida, we built an optic system for zooplankton research. That system could be towed through the water and look at all the organisms in the Gulf of Mexico and categorize them into fifty different groups for analysis. We have one of the guys from that project on our team. He’s one of the best in the world at writing that kind of software. That’s why our system works.
Background on Eric Steimle: I have a Ph.D. in Physical Chemistry from the University of South Florida. I’ve done a lot of particle size analyses, multi-wave length analyses and also designed sensors for particle counters. I’ve worked on liquid-core wave guides, which are similar to fiber optics. You put light in one end and you measure the spectrum that comes out the other end. Really cool stuff. When I finished my Ph.D., the oceanographic group at the University of South Florida asked if I could build stuff for them. So I started making equipment for them. That’s where I met Will. We’ve worked on oceanographic equipment for ten years before starting Larcos Aquaculture.
Information: John Cooksey, World Aquaculture Conference Management, P.O. Box 2302, Valley Center, California 92082, USA (phone 1-760-751-5005, fax 1-760-751-5003, email firstname.lastname@example.org, webpage https://www.was.org/Main/Default.asp).
Information: For a video of the Larcos PL-Counter in action, click on the following link http://www.larcosaqua.com/livestock_Tally_images.php.
Information: Eric Steimle, Ph.D., Program Manager (phone 1-727-492-8951, email email@example.com); Will Flanery, Embedded Systems Engineer (phone 1-727-804-5807, email firstname.lastname@example.org), and Kurt Kramer, Ph.D., Research Software Engineer (phone 1-813-263-4837, email email@example.com), Larcos Aquaculture (fax 1-727-851-9023, webpage http://www.larcosaqua.com).
Source: Eric Steimle and Will Flanery. Interview by Bob Rosenberry, Shrimp News International on March 2, 2012.