Observe’s state of the art use of Artificial Intelligence (AI) in aquaculture provides actionable insights in order to optimise the largest expenses facing fish farms.
Fish are one of the world’s largest sources of food. Roughly a third of the world rely on fish as their primary source of animal protein; and this is only set to increase as populations grow. Natural fishing reserves are already being depleted much more quickly than they can be regenerated, and as global demand increases it is clear that this model is simply not sustainable. Over the past two decades, global production from wild capture fisheries plateaued at approximately 90 million tonnes a year, yet the global fish supply has almost doubled to around 170 million tonnes. This was achieved thanks to growth in aquaculture, which now provides half of all fish for human consumption and is expected to continue to grow by 38.8% by 2025.
The benefits of aquaculture
Although less mature as a sector than land-based animal husbandry, aquaculture has far better feed conversion ratios (FCRs: the ratio of how many more kilogrammes of food the animal needs to eat to grow a kilogramme in weight) compared to typical livestock. The FCR for pigs and sheep is typically in the range 3kg to 6kg, while for beef it is 4.5kg to 7.5kg; whereas cultured finfish typically have an FCR of less than 1.5kg, with Atlantic salmon exhibiting an FCR of as little as 1kg to 1.1kg with efficient feeding. The principal reason for this is that cultured fish species are cold-blooded, meaning less of the energy they draw from food is wasted on heating their bodies. Moreover, fish protein is highly nutritious, rich in micronutrients, minerals and several vitamins, as well as omega-3 fatty acids, and represents an important part of a varied and healthy diet. As such, aquaculture represents a highly efficient and sustainable source of animal protein, and further developments in aquaculture best practice should be encouraged.
Salmon farming has expanded significantly over the past 30 years, resulting in most suitable sites already being utilised at capacity. Farmers now must consider improving productivity to increase output and cut costs; 32-43% of which count towards feed. Generally the most important task for farmers is to maximise feed output while minimising wasted feed. When feed is wasted, farm profitability, FCR, and local environment is all negatively impacted. Similarly, underfeeding slows farm growth.
Currently, aquaculture addresses this balance by relying heavily on human intuition and interpretations of existing data streams found on sites. Farmers spend their entire day monitoring their stock on video screens, looking for changes in fish behaviour to try and determine the fish’s satiety. They then manually adjust the amount of feed released into the cage based on what they see. This requires constant concentration, as farmers combine their interpretation of visual data with a plethora of environmental data streams gauging oxygen levels and current rates, feed intensities, historic fish growth and many more factors.
At the moment, there is no way of monitoring and evaluating feeding strategies objectively. Feeding strategies are an ‘art’: a reflection of the farmers’ spontaneous decision on how much food to put in a cage based on what they saw over the day. There are no unbiased measures of feed strategies to directly explain when a site produces good or bad FCR scores.
Introducing AI in aquaculture
Observe Technologies aims to convert aquaculture processes from an art to a science by using Artificial Intelligence and data processing to identify measurable patterns in feeding activities and strategies to present to farmers. The system is built to be adaptable and empowering for farmers by seamlessly tapping into the existing sensors, feed systems and cameras on site. We collect this data through a multitude of AI algorithms to optimise the farm’s performance; from the cost efficient use of feed to maintaining fish welfare. Past innovations have focused on hardware and data collection; however we discovered the problem is not a lack of data, but the rigour and overwhelming pressure on farmers to consistently interpret that data and apply correlations with fish activity, feeding patterns, sensory data, food particles and other historical information in real time. Our aim is to use AI in aquaculture to allow farmers to do more with the information already available on farms.
To facilitate this idea, Observe requires no new cage equipment to run on a site. Our product is a standalone ‘plug and play’ system which can interface with any existing camera streams found in salmon farms, analyse them in milliseconds to provide a standardised view of fish activity and the detection of food particles at different depths. Furthermore, companies have the opportunity to plug in their own site sensors, feeding systems and other auxiliary data to make the analysis more comprehensive leading to the automation of farms.
The use of AI in aquaculture farms offers constant analytical and objective evidence of how fish growth responds to farmer input under different conditions, allowing meaningful data exploration of different feeding strategies. As more sites are licensed, our AI systems leverage cloud infrastructures to provide remote site level analysis and anomaly detection to further expose farmers to data driven insights.
Global reach
We have made sure that farmers are the lead and at the heart of our development. To understand the problems faced by operators, we have operated AI in aquaculture on sites across multiple continents to develop not only an initial product, but a roadmap of development to ensure complete investment in the platform for the future. As a result, anecdotal feedback has been positive across the world.
Our technological talent is among the best in the world, and we have the ability to spread globally via our partnership with the AKVA Group ASA, the world’s largest aquaculture equipment provider, which has a fantastic approach and understanding of the sector. The AKVA group provides a level of on-ground support for farmers across the world, and valued expertise to develop more integrated and complete AI systems.
Aiding aquaculture’s sustainability efforts
The potential of AI in aquaculture does not simply end with local farm economics. The system can provide the potential to power data driven insights into aiding the environmental and sustainability drives of this sector. The Scottish parliament found that Scotland’s marine ecosystem is in jeopardy of ‘irrecoverable damage’ from salmon farming if environmental concerns are not addressed, and that Scotland is at ‘a critical point in considering how salmon farming develops in a sustainable way in relation to the environment’. Salmon farming output is largely limited primarily by the availability of licences, which is in turn limited by the environmental impact of salmon farming. Thus, the only practical measures to increase production are to minimise environmental impact whilst improving the efficiency of processes, which can be in part achieved by reducing the quantity of wasted feed.
Our AI algorithms can measure the poor feeding that contributes to damaged floor beds and provide real time insights before this occurs. In the future, it is our aim to provide industry compliance across the board to increase these sustainability efforts.
Just the beginning
Improving the environmental footprint and providing the possibility to standardise and reduce costs is just the start for AI in aquaculture. Going forward, we intend for our AI systems to automate entire aspects of the feeding process, bringing significant increases in productivity, and lowering costs for consumers.
Over the next few months, we will be introducing systems to measure the biomass of the fish automatically in the water as well as systematic automation. When scaled, the impact this will have on the blue economy and environment is immense. By utilising a cost effective way to feed, Observe aims to greatly reduce the environmental impact of aquaculture to help the industry continue to expand. It also allows us to better monitor the health of the cages, using anomaly detection to identify at risk sites and respond before they become an issue. This creates a new world of adherence to compliance over food quality whilst making a fish farmer’s life easier.
Additionally, though we only work with salmon at present, it is simple to apply the same architecture to all fish production. The adaptability of our learning algorithm means AI in aquaculture is incredibly transferable and will eventually allow us to branch out across the €178.34bn aquaculture market. Through closely working with farms we want to help them to analyse their data to provide benefits for them across the whole value chain, from feed companies to regulators.
Hemang Rishi
Co-Founder & CEO
Observe Technologies Ltd
+44 (0) 7968 949680
https://www.observe.tech/
