Identiq

• 

Inspiration

At almost every event (conferences, graduations, weddings, or festivals), hundreds or even thousands of photos are taken. After the event, these photos are usually shared through large folders on platforms like Google Drive or Dropbox. Attendees must scroll through hundreds of images just to find the few photos where they appear. Many people never find their photos at all. This made us ask a simple question: Why should people search for their photos, when technology could make photos find them automatically? What if Our personal Photos land into bad hands?

These question inspired Identiq, an AI platform that automatically detects attendees in event photos and delivers those photos directly to them.

Our goal is to make event memories easier to access, more personal, instantly shareable, and most importantly more secure.

What it does

Identiq is an AI event photo delivery platform. It automatically identifies attendees in event photos and sends them the photos where they appear. The workflow is simple: 1- At the event entrance, attendees scan a QR code. 2- They take a selfie and enter their email. 3- Photographers upload event photos to the Identiq platform. 4- Our AI analyzes each photo and detects faces. 5- The system matches detected faces with registered attendees. 6- Each attendee receives only the photos where they appear. Even if the attendee appears: alone, in a group photo, partially in the background

Identiq finds them and delivers their photos automatically.

How we built it

Identiq combines modern AI, cloud infrastructure, and web technologies to automate event photo distribution.

Frontend

We built the user interface using:

Next.js Tailwind CSS

The frontend provides:

Organizer dashboard Event creation Photo upload interface QR code generation for attendees Attendee registration page Private photo galleries

Backend

The backend was built with:

• FastAPI (Python)

It handles:

attendee registration I face processing photo matching automated email delivery

AI Pipeline

Our AI system works in two stages:

1. Face Detection

We use YOLO (You Only Look Once) models to detect faces in event photos.

This allows us to efficiently identify multiple faces in large images.

2. Face Recognition

We use DeepFace embeddings to represent each face as a numerical vector.

Each attendee selfie produces a face embedding:

[ E_{attendee} \in \mathbb{R}^{n} ]

Each detected face in an event photo produces another embedding:

[ E_{photo} \in \mathbb{R}^{n} ]

We then compute similarity using cosine similarity:

[ similarity = \frac{E_{attendee} \cdot E_{photo}}{|E_{attendee}| |E_{photo}|} ]

If the similarity score passes a threshold, the system considers it a match.

Database & Infrastructure

We used Supabase for:

PostgreSQL database storage for event photos authentication row-level security

Images are stored in Supabase Storage and processed through our AI pipeline.

Challenges we ran into

Building Identiq involved several technical and design challenges.

1. Face Matching Accuracy

Detecting faces in event photos is difficult because:

lighting conditions vary people appear in different angles group photos contain multiple faces

We addressed this by combining YOLO detection with DeepFace embeddings and similarity thresholds.

2. Performance

Events can contain:

hundreds of attendees thousands of photos

A naive approach would require comparing every face against every attendee, which would be slow.

To improve efficiency, we designed the pipeline to process matches attendee-by-attendee and optimized the comparison workflow.

3. Storage and Security

Event photos contain sensitive data.

We used Supabase Row-Level Security (RLS) to ensure:

organizers only access their own events attendees only receive photos matched to them storage access is controlled

Accomplishments that we're proud of

We are proud that we built a complete end-to-end system during the hackathon.

Our platform includes:

attendee registration via QR code or a link AI-based face detection and matching automated event photo distribution organizer dashboard attendee photo galleries

The system demonstrates how AI can transform a manual and frustrating process into a fully automated experience.

Most importantly, we created a solution that solves a real problem experienced at almost every event.

What we learned

Building Identiq taught us several valuable lessons.

AI in Real-World Systems

Implementing AI is not just about models, it requires building a full pipeline that integrates:

detection feature extraction similarity search system performance

Scalability Matters

Handling large numbers of photos requires careful design to avoid inefficient comparisons.

User Experience is Critical

Even the best AI system must be wrapped in a simple, intuitive experience.

The QR-code-based attendee flow made the system accessible and easy to use.

What's next for Identiq

Our vision is to evolve Identiq into a full SaaS platform for event photography.

Next steps include:

1. Faster Face Matching

Integrate FAISS vector search to scale matching to large events with thousands of attendees.

2. Mobile App

Allow attendees to:

receive instant notifications browse event galleries download and share photos easily

3. Photographer Tools

Provide professional photographers with:

automated photo delivery client management event analytics.

4. Privacy Controls

Implement stronger privacy features, including:

• opt-in face recognition
• secure photo access
• user-controlled data deletion.

5. Large Event Support

Scale the platform to support:

conferences festivals stadium events universities

with thousands of participants.

Built With

• deepface
• nextjs
• python
• yolo