Curtin Institute for Data ScienceFish identification
Fish in underwater footage were manually identified and counted.
Collaboration
Harrier Project Management · SeaGIS · In-Situ Marine Optics · Business Research and Innovation Initiative (BRII)
Automated Fish Identification (AFID)
AFID uses computer vision to automate fish detection, identification, and length measurement from underwater video, making fisheries monitoring faster and more scalable.
- Australian organisations
- 7
- International partners
- 3
- Peer-reviewed publications
- 2
Project overview
Turning underwater video into automated fisheries insights.
Australia’s fisheries industry is valued at more than $2.5 billion annually, making effective monitoring of fish populations important for sustainable fisheries management. Traditional analysis of Baited Remote Underwater Video Systems (BRUVS) footage is labour intensive, time consuming and costly.
CIDS worked with marine technology and ecological monitoring partners to develop the Automated Fish Identification (AFID) system. The solution uses computer vision and deep learning to detect and identify fish species, measure fish length and integrate AI-generated results into EventMeasure, an established video analysis platform.
The resulting proof of concept included fish detection, automated length measurement, species identification, fish tracking, MaxN discovery, bulk labelling and data management capabilities. Testing with users helped refine the algorithms and demonstrated significant time savings compared with manual annotation.
The web-based system provides a pathway towards more efficient and scalable fisheries monitoring and can be licensed for adoption by organisations in the sector.
Before and after
3 operational shifts from the project
Before
After
Fish identification
Machine learning automatically detects and identifies fish species.
Reduces manual annotation effort and makes video analysis more efficient.
Video analysis
BRUVS footage required extensive manual review and annotation.
Video analysis
AI outputs are integrated into EventMeasure alongside existing analysis workflows.
Reduces labour-intensive processing and enables researchers to analyse footage more efficiently.
Fisheries monitoring
The time and cost of manual analysis limited the scale of underwater video monitoring.
Fisheries monitoring
AFID provides a web-based solution for automated fisheries analysis.
Creates a pathway to faster, more scalable and potentially licensable fisheries monitoring.
Have a challenge? Bring it to CIDS.
We bring together the expertise, technology and research capability to determine the right way forward.