How does it work?

NeuralFin uses biometric pattern recognition similar to facial recognition in humans to identify individual sharks from images. When an image is uploaded, the system detects key markers such as dorsal fin shape, spot patterns, and scarring. It also analyses subtle signals—micro-textures, curvature, and contrast patterns across multiple spatial frequencies often beyond what the unaided human eye can reliably distinguish. These features are converted into a unique numerical fingerprint and matched against our global database using deep learning algorithms.

If the shark has been seen before, its existing ID is returned with metadata including previous sightings. If not, NeuralFin assigns a brand-new ID and archives the image and pattern data for future recognition.

This process, while fully automated, is rooted in collaboration with marine scientists to ensure species-specific accuracy. NeuralFin is continuously updated and trained on new images to improve performance over time.

Shark ID AI has been tested with university partners and is in active use by PhD candidates as part of their research. Projects use the platform to generate publishable evidence on mark–recapture, site fidelity and movement, with methods and data checks reviewed by academic supervisors. Records include stable individual IDs with date and location, so outputs align with standard research requirements.