jeba-jebarsan/ai-event-photo-finder

AI Event Photo Finder solves a problem anyone who has been to a wedding or conference knows well: a photographer takes hundreds of photos, and finding the ones you're actually in means scrolling through all of them. This project lets a photographer upload every photo from an event once, and then lets each guest instantly get back only their own photos. Here is how it works. A photographer creates an account, sets up an event, and uploads the full set of photos. The system then scans every photo, detects each face in it, and creates a mathematical representation of each face on the server. The event gets its own QR code, which the photographer shares with guests, for example by printing it or displaying it at the venue. A guest scans the code, takes a live selfie through a guided capture process that checks the photo quality, and within seconds gets back the photos from that event where their face was matched, sorted by how confident the match is. No app install or account is needed on the guest side. The project is split into two parts. The web app is built with Next.js and TypeScript, styled with TailwindCSS, and handles accounts, event creation, uploads, QR codes, and the guest facing flow. The face detection and matching is a separate Python service built with FastAPI, using an open source face recognition library called InsightFace. Supabase provides the database, authentication, and file storage behind the scenes. According to the project's status notes, the core flow from uploading photos to getting matched results back works end to end, but the project has not been deployed publicly yet. Known open items include moving to a more scalable way of storing face data for handling many events at once, notifying guests when their photos are ready, figuring out pricing, and writing a formal privacy policy for handling biometric face data, which the maintainers flag as still missing. The project is released under the MIT license and welcomes outside contributions, while noting it is early stage and may have rough edges.