How Artificial Intelligence is Transforming the Aquaculture Sector

Salmon cages. Australia's aquaculture industry has an annual value of more than 00 million and is expanding at the rate of 20% per year. CSIRO is the country's largest supporter of aquaculture research, injecting .5 million into the industry each year. (Credit: CSIRO via Wikimedia Commons CC BY 3.0)

Leveraging technological advancements, like the boom and rapid growth of artificial intelligence (AI), is essential to ensuring that aquaculture operations maximize production and sustainability. Advancements in AI-driven monitoring, fish feeding, screening, and product analysis are fueled by deep learning, machine vision, and Internet of Things (IoT) technology.

A 2023 review of current applications of AI technology in aquaculture summarizes several key uses that have been applied in operations across the world. The advancements discussed are focused on the last half-decade, and the review highlights opportunities for growth in the future.

Monitoring Environmental Conditions using AI

Monitoring environmental conditions, such as water quality, is essential to informing aquaculture operations. For example, if water quality dips unexpectedly, a real-time, continuous monitoring system can alert operators of the dip, allowing them to respond quickly and accordingly with hydrogen peroxide dosing or turning on aerators. 

Smart aquaculture monitoring, such as this, improves upon traditional approaches that measure water quality conditions on a fixed schedule, which could result in missing hypoxic events or other stressors that result in product losses. Popular approaches include data buoys, unmanned aerial vehicles (UAVs), or stationary pole-mount systems.

Both data buoys and stationary pole-mount systems offer the ability to measure water quality continuously, though they are limited spatially as data is collected at a fixed point where the system is deployed. However, these stationary systems, when equipped with telemetric data loggers, offer high-quality data in real-time at a lower power requirement. 

For operators interested in higher spatial resolution, UAVs collect data over a larger area. AI drones use satellite remote sensing to extrapolate electrical conductivity, water depth, turbidity, pH, temperature, nitrates, chlorophyll, and dissolved oxygen measurements. 

UAVs can be more power-hungry than alternative solutions, which the review notes could result in shorter flights that may not cover the entire farm–however, future battery advancements may resolve this issue.

Ultimately, stationary options like data buoys and pole-mount systems offer a more continuous data collection option as these systems collect data more frequently than UAVs, which have planned, scheduled flights that are predetermined. In short, stationary environmental monitoring approaches offer higher temporal resolution while UAVs observe a larger area during flights, offering improved spatial resolution.

Smart Feeding

According to the review, feed for aquaculture operations accounts for around 60% of aquaculture costs, making overfeeding a costly mistake accompanied by water quality declines. In contrast, underfeeding can negatively impact growth, health, and behavior. As such, smart feeding technology is a recent innovation that gauges hunger and times feeding appropriately.

Per the review, “The acoustic signals and vibration-based sensor are capable of distinguishing hungry fish from well-fed ones.” When paired with a smart feeder, the entire process becomes automated by AI technology, eliminating the need for manual intervention.

Smart feeding technology helps minimize overfeeding, which can deplete water quality due to decomposing feed, as well as prevent underfeeding, which can cause diminishing muscle conversion and cannibalism. 

Circle floating aquaculture fish farming cage. (Credit: Luc Coekaerts via Wikimedia Commons CC0 1.0)

Smart Fish Screening, Harvesting, and Processing

Fish growth can be influenced by more than just feeding techniques, linked also to water quality. As such, fish seed screening is essential to ensure that seeded fish are able to grow into healthy adults (or the age at which they are harvested). 

While traditional screening approaches are laborious and costly, the review notes a smart application of Microsoft Azure Machine Learning Studio with IoT and other AI technologies to sort fingerlings based on size and viability. This automated sorting helps separate fish that are less likely to survive.

Once separated, automated systems may allow the small and unwanted fish to exit the captive system safely or separate sick fish to another holding facility to prevent the spread of disease amongst the harvest crop. 

Once harvested, AI innovations like vision technology can help automate and speed up processing. Improved processing speeds and maximizing harvest population help improve efficiency.

Behavior Analysis and Disease Diagnosis through Artificial Intelligence

The review highlights various AI approaches to analyze behaviors, including support vector machine (SVM) and deep convolutional neural networks (DCNNs). Such methods can be used to determine sex based on caudal fin color and appearance texture of the fish.

Another approach highlighted in the review includes the use of computer vision and deep learning to conduct aquatic toxic analysis by monitoring fish behavior. Additionally, a novel analysis of hydroacoustic datasets employs automated machine learning tools.

According to the review, “One major advantage of this approach is that it allows high-quality models that are specific to the data at hand to be trained even in the absence of data science expertise.” 

Overall, these approaches enable the high-density measurement of fish behavior, lice count, gill status, and mortalities using image-based machine learning and other AI technologies.

Intelligent Aquaculture

As the industry continues to grow, AI technology will become integral to maximizing the production of aquaculture operations. Innovation paired with sustainability is essential to ensuring the industry continues to grow and meet future food needs.

Read the full review here.