In shrimp farming, operators must regularly remove the animals from the water to weigh them and check their condition. However, this causes stress and reduces animal welfare. It is also almost impossible to detect symptoms of stress or disease, as turbidity, even under optimal light conditions, often limits in-situ observation. This is where the ShrimpWiz project comes in: Led by the Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research (AWI), a team of scientists and engineers in collaboration with Oceanloop, a pioneer in European indoor shrimp farming, has developed a system that can count shrimp in images using AI-based computer vision software. Under realistic farming conditions and in real-time, the system can also determine the length of the animals with 95% accuracy.
The first prototype has been tested at Oceanloop's research and development farm in Kiel, Germany. An advanced smartphone installed above the water surface automatically photographs the shrimp once a minute and transmits the data to a local server. Here, computer vision algorithms count individual shrimp in each image and measure their length. By improving the image quality and using the latest generation of AI-based image processing models, the team was even able to detect visual signs of stress in the animals.
Unlike pond farming, Oceanloop's systems use clear water for farming. These systems are therefore ideal for computer vision, as the consortium demonstrated in the previous MonitorShrimp project. Due to the high turbidity of the water in traditional pond systems, it is virtually impossible to visually monitor the welfare of the animals, either with the naked eye or with computer vision.
Stephan Ende, coordinator of the project at the AWI, is convinced that clear water technology is the key to animal welfare in intensive aquaculture. "The use of computer vision software to measure shrimp enables accurate and non-invasive monitoring of animal welfare and productivity in shrimp farming - 24 hours a day, 7 days a week. The clear water technology combined with our Early Welfare Alert software can be the starting point for more objective welfare labeling in the shrimp industry of tomorrow".
The aim of ShrimpWiz is to develop market-ready computer vision hardware and software for indoor shrimp farming that can capture all the necessary information in a single image, including biomass, stress and - at a later stage - possible diseases.
"Non-invasive, real-time monitoring of key farming parameters such as growth, feed conversion, survival and stress will make a crucial contribution to a better understanding of shrimp farming. To better understand the needs of our shrimp, we can use these economically important indicators to develop an artificial neural network that takes into account all available farm data, which can easily add up to more than a hundred," said Bert Wecker, CTO of Oceanloop.
Tomasz Kowalczyk, founder and CEO of NeuroSYS, which was involved in developing the algorithm for the project, explained that “technological advances can transform companies and entire industries. We are ready to be part of this change and are working to bring the benefits of artificial intelligence and deep learning to the shrimp farming industry.”
The consortium sees the development of AI-based software as an opportunity not only to improve animal welfare but also to increase farming efficiency. The technology can help drive the digitalization of indoor shrimp farming, which is necessary to achieve today's retail price levels. "Proving the technical feasibility of alternative solutions is crucial to meet the growing awareness of customers and stakeholders for more sustainable and welfare-compliant shrimp farming," concluded Stephan Ende.
The project is funded by the Federal Ministry of Food and Agriculture (BMEL) on the basis of a decision by the German Bundestag via the Federal Office for Agriculture and Food (BLE) as part of the innovation funding program.
