dr snapper - capturing fish on camera
For all folks interested in fish, cameras and AI, this will be the place to meet and mingle in March this year: 2nd international Symposium on Catch Identification Technologies. En-join in Tromsø, Norway, 19th-20th March 2024.
Picture: AI-generated by Adobe Photoshop
Caring for fish was the central theme of the catch welfare platform’s kick-off meeting in Bergen, Noway last week. While the majority of presentations were about how to improve treatments of fish in wild-capture commercial fisheries, the pros and cons and welfare challenges of recreational, catch and release were neatly summarized as well.
To present results from the European smartfish project, a mini-symposium was organized at Norway House (the Norwegian embassy) in Brussels last week. What was on offer: some of the latest high tech for commercial fishing activities to optimize their resource efficiency, improve automatic data collection, monitoring compliance to overall reduce the ecological footprint of the extraction industry. Some hardware that was on display included: catch scanners - cameras that register and identify (via AI) catches on conveyor belts, underwater cameras that monitor what enters into a net to allow for real-time decisions whether to continue fishing or not (see an earlier post when I joined a field trial in Denmark). Acoustic triggers (from marport) to a receiver with a release mechanism to close a codend underwater. This can come in handy for actively manipulating catch compositions. For example, a skipper can decide to close the codend as soon as the target species enters the net, thereby reducing the amount of unwanted bycatch. Visual light sensors to illuminate parts of fishing gear to either attract target or deter unwanted bycatch species to eventually improve trawl selectivity. if you are interested in more detailed results of how these technologies can make fishing more sustainable, feel free to get in touch and I can connect you with the involved researchers and technology experts.
TrawlMonitor camera
Underwater camera system for real-time images from inside the fishing gear.
Electronic monitoring
AI-based species recognition and count of catch compositions.
Overview of technological advancements
Hard- and software products forthcoming from the smartfish project.
The ASFB is an exciting conference for anyone interested in fish and fisheries. The keynote of day 1 sounds like a cracker (see abstract below) - it will heavily feature the pioneering work of the FishID platform - I held some talks with one of their scientists last year - hopefully a collaboration will eventuate in the not so distant future:
Rod Connolly 1
Griffith University, Ashmore, QLD, Australia
The revolutionary use of automation combining computer vision and big data analytics is finally coming in to land in fisheries science! The presentation will explain opportunities and challenges that researchers and managers face in automated monitoring of fish abundance, biomass, and behaviour. Computer vision on robust and inexpensive camera systems can be particularly valuable for increasing efficiency and reliability in monitoring that is currently difficult, dangerous or prohibitively expensive. It can supercharge science. It can improve management – of wild fisheries and aquaculture. Increased automation is helping in stock assessments, and in monitoring to detect trends in the abundance and biomass of fish and benthic animals. Computer vision is also enabling more efficient monitoring of the extent and condition of fish habitat, of the presence of invasive species, and of interactions among species. New analytics are being developed as we move from the traditional scenario of having samples sizes that are often too small to be reliable, to a scenario of having far more data than fits on a hard-drive! So automation is creating new training needs and opportunities. There is a need for mentoring of graduates in statistics, for QA-QC, and for coping with embedded pseudo-replication. And always there is the imperative to sharpen our focus on outputs that inform actions.
