SkateSpot

Skate Spots
Basic Map Implementation
Hotspots
Information about Spot
Traffic Evaluation
AI Rating
In depth AI analysis
Re-evaluate function
OAuth Screen

Inspiration

We talked to representative of TAMU Skates. He said TAMU skaters were having problems finding good spots to skate on campus. He also said it was hard for new skaters to get into skating and join a community

What it does

Google OAuth lets users sign in with their Google account, giving us a verified identity (name, email, avatar) without storing passwords Click anywhere on the map to add a pin Add the spot’s name, description, and image After adding, the pin will be accessible to anyone who’s using the app Users have opportunities to get an AI analysis of the skate spot whenever they upload pictures It will give the total rating of the skating spot and AI analysis of why it is a good/bad spot to skate The Heat Map shows the crowd density The hotter (red) area means more people Shows the number of people at the spot

How we built it

Frontend: Expo + React Native + Mapbox Backend (Supabase) : Auth with google sign in; Postgres tables for spots, spot photos, and skate spot rating; and Google Cloud vision API to perform Image Analysis.

Challenges we ran into

Native Mapbox (@rnmapbox/maps) means Expo Go or web previews can’t render the map; teammates need the custom dev client or full native build. iOS uses Mapbox’s static frameworks—any Pods still expecting dynamic frameworks will cause link-time build failures. OAuth setup: Web implicit puts tokens in URL fragments; mobile deep links drop fragments. use Code + PKCE. No shared browser cookies on mobile; store app tokens locally for the session. Google Vision API: Keep Vision keys off the app. Upload the photo to Supabase Storage, then have an Edge Function call Vision and save the score back to the DB. Don’t send file:// paths. Upload first and pass bytes or a signed URL. Shrink images (~1280px, ~70% quality) so it’s fast and cheap.

Accomplishments that we're proud of

A working MVP: Add/view spots with photos instant map pins, and an AI rating + explanation

What we learned

How to ship native Mapbox inside Expo Dev Builds The value of structured spot data for making app

What's next for SkateSpot

Training an AI model to handle different image types and give more accurate rating of the skating spots Searching with keywords such as smoothness or stairs