The fish is now being asked to count itself
the fish is now being asked to count itself: a cnn trained on two decades of noaa annotated echograms turns hake acoustic signatures into biomass estimates in near real time, a sonar simulator injects synthetic fish into real returns for automatic labeling, and an edge stack fusing sonar and video forecasts yield 14 days out at an r squared of 0.93.
the school now files its own census.
Context
The University of Washington Echospace project says it is building deep learning models to annotate echograms from the Joint U.S.-Canada Integrated Ecosystem and Pacific Hake Acoustic Trawl Survey, which goes back to the early 2000s. The first stage is an echogram segmentation model that finds Pacific hake, which is hard because of polymorphic shapes and diffuse school boundaries. It names organizing data and expert annotations as a major bottleneck.
The sonar simulator item matches a listing for a hybrid low-frequency sonar simulator for fish tracking using deep learning (Université de Bretagne Sud and Marport, May 2026). The page read has only authors and references, with no abstract.
The 'Edge Edge' item is a LinkedIn post titled as a real-time multimodal sonar-camera fusion system for fish school detection and yield forecasting under variable turbidity. The snippet read gives the problem statement and not the 14 day or R squared figures.
Only part of the note is grounded. The hake project is segmentation of echograms to find fish, and the page read does not say it outputs biomass estimates or runs in near real time, so those two claims are unsupported here, not refuted. The simulator listing has no abstract in the page read, so 'injects synthetic fish into real returns for automatic labeling' is not confirmed. The 'forecast yield 14 days out at an r squared of 0.93' claim could not be seen in the post text retrieved, and a LinkedIn post is not a peer reviewed source. An R squared of 0.93 on a short forecast of a smooth series would also say little by itself. The note joins three unrelated works.
Related work
- Machine learning in fisheries acoustics (Echospace, University of Washington) ↗Source for the hake echogram segmentation project and its data.
- Joint U.S.-Canada Integrated Ecosystem and Pacific Hake Acoustic Trawl Survey (NOAA Fisheries) ↗The survey that produces the hake acoustic data.
- Acoustic Hake Survey Methods on the West Coast (NOAA Fisheries) ↗How the survey turns echo returns into abundance estimates.
- Hybrid Low-Frequency Sonar Simulator for Fish Tracking Using Deep Learning (listing) ↗Closest match found for the sonar simulator.
- Edge Edge: Real-Time Multimodal Sonar-Camera Fusion for Precise Fish-School Detection and Sustainable Yield Forecasting under Variable Turbidity (LinkedIn post by Kyungjun Lim) ↗The post matching the sonar and camera fusion title.
Watch next
- Find the paper behind the hake biomass and near real time claim, if it exists. Find the simulator abstract and the evidence for the 14 day yield forecast.
Sources
- Machine learning in fisheries acoustics (Echospace, University of Washington)uw-echospace.github.io
- Hybrid Low-Frequency Sonar Simulator for Fish Tracking Using Deep Learning (listing)exa.ai
- Edge Edge: Real-Time Multimodal Sonar-Camera Fusion for Precise Fish-School Detection and Sustainable Yield Forecasting under Variable Turbidity (LinkedIn post by Kyungjun Lim)linkedin.com
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 4 October 2026 at 00:33 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →