FISHDEX: Fish ID by Photo

Inspiration

Fish tagging and invasive marking methods cause stress and physical harm to fish, and they are costly and time-consuming for researchers, hatcheries, and anglers. Recent advances in non-invasive photo-identification research (using natural skin patterns, scale structures, and lateral line shapes) showed that fish individuals and species can be reliably recognized from a simple photograph. We wanted to bring that same non-invasive identification power into an accessible mobile app that anyone could use in the field.

What it does

FISHDEX lets users snap or upload a photo of a fish and instantly get a species identification based on its visual features, no tagging, handling, or specialized equipment required. It's designed for anglers who want to log their catches, researchers doing biodiversity surveys, and conservationists monitoring native and invasive species populations, all through a fast, non-invasive visual workflow.

How we built it

We combined a mobile-first app experience with an AI-powered image classification pipeline. Photos submitted by users are processed through a computer vision model trained to recognize distinguishing visual features of fish species (body shape, coloration, fin structure, and patterning). We integrated RevenueCat for subscription and monetization management, allowing us to offer premium identification and logging features while keeping the core experience accessible.

Challenges we ran into

Building a robust visual identification system that works across varying lighting conditions, angles, and partial occlusions in real-world field photos was the biggest technical hurdle. We also had to balance model accuracy with performance on mobile devices, and carefully design the monetization flow so it didn't get in the way of the core non-invasive identification experience.

Accomplishments that we're proud of

We built a working end-to-end pipeline that takes a photo from a user's phone all the way to a species identification result, integrated with a smooth subscription experience via RevenueCat.

What we learned

We learned a great deal about applying computer vision techniques to biological identification problems, and about integrating in-app purchase and subscription infrastructure using RevenueCat within a mobile hackathon timeframe.

What's next for FISHDEX

We plan to expand our species database, improve model accuracy across more diverse aquatic environments, and add community features so users can contribute verified sightings to help track biodiversity and invasive species over time.

Built With

• computer-vision
• firebase
• python
• react-native
• revenuecat
• tensorflow
• typescript