Live Photoplankton Analysis System (LPAS)
SPECIAL THANKS: To all companies that have assisted with training the artificial intelligence (AI). Further image collection has also been made possible through laboratory partners in the UK & Ireland, Canada, Chile and Australia.
COLLABORATORS
FUNDERS
-
To develop and provide automatic identification of harfum algal bloom phytoplankton species. This on-site system will enable increased testing frequency, more rapid turn-around time, and improved consistency in species identification. The device and software solution will consists of:
User defined alerts warning of presence of specific species and concentration levels
Local and Cloud access to data
Potentially generate data set for long term analysis/prediction
Continuous training and update of AI engine.
There are a number of challenges with the current processes of identifying HAB species:
Identification methods are often labour intensive which impacts how rapidly decisions can be made around feed management and other mitigation methods
It is difficult to determine genus or species
Other testing methods that require off-site laboratories increase turn-around time
-
Collaborate with farms to collect imagery in order to train the artificial intelligence engine
Once the AI is trained, then there will be a period of field-testing and pre-commercial implementation
Develop the solution for both saltwater and freshwater algal species
-
We deployed the devices on farms in Australia (salmon, prawns and barramundi) and tutorials were provided to companies. The methods used typically fits into operational processes. Images of algae taken by the farms are identified by the farms or the LPAS experts via uploading images to the cloud. To date the LPAS data collection phase has received>4,750 images of phytoplankton from wild samples taken at aquaculture sites (Figure 1). This brings the total number of images to >9500 when combined with laboratory-based data collection.
-
“real-time” and accurate data
Enables faster decision-making and implementation of mitigation measures
Improved fish welfare outcomes for farmed aquaculture species susceptible to HABs
-
BELINDA YAXLEY
The device is in pre-commercial stage on Scottish salmon farms and Tasmania is only two years behind having a commercial device available.
FIGURE 1: Total number of images taken by collaborating aquaculture sites in Chile, Ireland, Scotland, and Tasmania, as of 11/11/2022. An introductory list of HAB phytoplankton organisms have been selected for the initial release of the LPAS system to market (Table 1). Table 1. List of phytoplankton detectable by LPAS A.I. in phase 1.
*The Chaetoceros genus is divided into three groupings: Phaeoceros; large, harmful species with chloroplasts in the setae (C. convolutes, C. concavicornis), Hyalochaetae; smaller, less harmful species with clear setae (C.mitra, C. didymus), and C. socialis. These phytoplankton have been selected based on the data that has been received by the LPAS image collection phase, combined with input from industry experts and personnel. These phytoplankton will be detectable by the LPAS A.I. to a minimum quality threshold of accuracy by the time of release to the market.