SwipeLab

A Gamified Human-in-the-Loop Image Labeling Platform

SwipeLab is a cross-platform image labeling system designed to help researchers quickly collect high-quality human labels for image datasets. The platform includes a mobile app for users to classify images using a simple “Yes / No / I Don’t Know” interface, a secure backend for data storage and API communication, and a research dashboard for monitoring label quality and exporting results.

---

📸 Project Overview

Machine learning models require large, accurately labeled datasets. Many classification tasks still depend on human intuition, especially when images involve subtle patterns, aesthetics, or subjective interpretation.

SwipeLab provides a fast, intuitive, and gamified labeling interface that makes it easy for casual users and volunteers to contribute. The system ensures data quality through credibility scoring, gold-standard items, and progress tracking.

---

🧩 System Components

SwipeLab/
│
├── frontend/ # React Native (Expo) mobile app
│ └── swipe-based classification UI
│
├── backend/ # Java Spring Boot server
│ ├── REST API for images, labels, authentication
│ ├── PostgreSQL database
│ └── researcher dashboard & admin endpoints
│
└── docs/ # Architecture diagrams, proposal, design docs

---

🎯 Goals of the Project

• Provide researchers with a simple & scalable tool for collecting human-labeled data.
• Make labeling fast, intuitive, and accessible on any mobile device.
• Support human quality control with gold-standard items and credibility scoring.
• Deliver a secure backend with auditing, authentication, and export tools.
• Build a gamified system that motivates users to contribute more labels.

---

📱 Frontend (React Native + Expo)

The mobile app allows users to:

• Swipe through images to classify them
• Choose Yes / No / I Don’t Know with one tap
• Track personal progress (points, badges, streaks)
• Compete on leaderboards
• Rate confidence and improve credibility score
• View session history & statistics

Designed for:
✔ simplicity
✔ speed
✔ mobile-first
✔ minimal cognitive load

---

🔧 Backend (Spring Boot + PostgreSQL)

The backend provides:

• Secure REST API (HTTPS + OAuth2 Google Sign-In)
• Image batch retrieval
• Label submission & validation
• Gold image handling for quality control
• User accounts, roles, and credibility scoring
• Researcher dashboard with:
  • statistics & graphs
  • dataset export (CSV/JSON)
  • consensus levels
  • user analytics
  • error logs & audit trail

---

🧪 Quality & Testing

• Unit Tests: frontend components & backend services
• Integration Tests: app ↔ backend ↔ dataset API
• API Tests: schema validation, authentication, rate limits
• UI/UX Tests: usability and user studies with volunteers
• Performance Tests: response times, batch processing
• User Acceptance Testing (UAT): final evaluation by real researchers

---

🔐 Security Requirements

• HTTPS-only
• OAuth2 Google sign-in
• Token validation
• Short-lived signed URLs for images
• CORS & client restrictions
• Auditable logs (user, timestamp, label, elapsed time)
• Role-based access (user vs researcher/admin)

---

📚 Technologies

Frontend

• React Native
• Expo CLI
• TypeScript
• React Navigation
• Reanimated & Gesture Handler

Backend

• Spring Boot
• Spring Security + OAuth2
• Maven / Gradle
• PostgreSQL
• Docker
• AWS / GCP / Render

Tools

• GitHub + GitHub Actions CI/CD
• pgAdmin
• Postman
• draw.io for architecture diagrams

---

🚀 Getting Started

Clone repo

git clone https://github.com/edenbar23/SwipeLab.git

Setup Frontend

cd frontend
npm install
npm start

Setup Backend

cd ../backend
./mvnw spring-boot:run

👥 Team

Sagi Evroni Eden Bar Ofri Hanochi Avihoo Amos

Customer:

Prof. Chen Keasar Department of Interdisciplinary Computation Ben-Gurion University

📄 License

TBD — will be added before public release.