🎽 Reward Closet

AI-Powered Clothing Donation Platform

A web application that analyzes clothing condition using AI to predict reward points and enables easy clothing donations.

📖 Introduction

Project Overview

Reward Closet is an innovative platform that bridges the gap between clothing waste and charitable giving through artificial intelligence. Our mission is to make clothing donations more accessible, transparent, and rewarding for everyone involved.

The Problem We're Solving

Clothing Waste Crisis: Millions of tons of clothing end up in landfills each year, contributing to environmental pollution
Donation Barriers: Traditional donation processes are often complicated, time-consuming, and lack transparency
Value Uncertainty: Donors have no clear understanding of their clothing's worth or impact
Limited Incentives: Current donation systems provide minimal motivation for regular participation

Our Motivation

We believe that technology can transform how we approach sustainable consumption and charitable giving. By combining AI-powered assessment with a transparent reward system, we aim to:

Reduce Textile Waste: Make it easier for people to donate instead of discarding
Increase Donation Participation: Provide clear value and rewards to encourage regular donations
Enhance Transparency: Use technology to show donors the real impact of their contributions
Build Community: Create a platform where giving back becomes engaging and social

Our Solution

Reward Closet leverages cutting-edge AI technology to:

Instantly Assess: AI analyzes clothing condition and provides immediate value estimates
Reward Generosity: Points-based system that recognizes and rewards donors
Simplify Process: One-click donation system with flexible pickup options
Track Impact: Real-time tracking of donation status and community impact

By making donations rewarding, transparent, and effortless, we're creating a sustainable ecosystem where everyone wins – donors feel valued, recipients get quality clothing, and the environment benefits from reduced waste.

🔗 Project Links

🌐 Web Application: https://reward-closet.vercel.app
⚡ AI API Server Repository: https://github.com/abjin/reward-closet-ai-api-server
📚 AI API Documentation: http://reward-closet-ai-api.eba-by6v3rd4.ap-northeast-2.elasticbeanstalk.com/docs
📺 Youtube Video: https://youtu.be/X4Cu6lpv5bM

🏗 Representative block diagram (System Architecture)

The diagram above illustrates the core components and data flow of the Reward Closet platform:

User: Interacts with the web application
Web App: Next.js application hosted on Vercel
AI Server: FastAPI-based clothing analysis service on AWS
Database: MySQL database for user data and donation records
Supabase: Handles authentication and image storage

✨ Key Features

🤖 AI Clothing Condition Prediction

AI analyzes uploaded clothing photos to evaluate condition
Provides estimated reward points based on clothing condition
Automatic analysis of item type, and condition

💝 Simple Donation System

One-click donation application
Delivery pickup and home collection options
Real-time donation progress tracking

🏆 Point Reward System

Points awarded upon donation completion
Personal accumulated points tracking
Donation history management

👤 User Management

Supabase Auth based secure registration/login
Email verification and social login support
Personal donation history and points management
Complete activity overview in My Page

💾 File Management

Image upload using Supabase Storage
Automatic file size and format validation (max 10MB, JPG/PNG/WebP)
Fast image serving through CDN

🛠 Tech Stack

Artificial Intelligence

Python - Base framework
YOLO - Special library & models for computer vision

Frontend

Next.js 15 - React-based full-stack framework
React 19 - Latest React version
TypeScript - Type safety
Tailwind CSS - Utility-first CSS framework
Radix UI - Accessible UI components
Lucide React - Icon library

Backend & Database

Next.js API Routes - Serverless API
Prisma - Type-safe ORM
MySQL - Relational database
Supabase Auth - Authentication system
Supabase Storage - Image file storage

Deployment & Infrastructure

Vercel - Next.js optimized hosting platform
Supabase - Auth, Storage, real-time features
AWS Elastic Beanstalk - AI API service hosting

Development Tools

ESLint - Code quality management
Turbopack - Fast development server

📁 Project Structure

data/
├── preprocessing.ipynb  # Data preprocessing
└── clothes.yaml         # YOLO formatted data file

prisma/
├── schema.prisma        # Database schema
└── migrations/          # Database migrations

src/
├── app/                 # Next.js App Router pages
│   ├── page.tsx         # Home page
│   ├── login/           # Login page
│   ├── signup/          # Sign up page
│   ├── predict/         # AI prediction page
│   ├── donate/          # Donation application page
│   ├── mypage/          # My page
│   ├── guide/           # Donation guide
│   └── api/             # API routes
│       ├── users/       # User-related API
│       ├── donations/   # Donation-related API
│       ├── predict/     # AI prediction API
│       └── user/        # Individual user API
├── components/          # Reusable components
│   ├── ui/              # Basic UI components
│   └── header.tsx       # Common header
├── lib/                 # Utilities and configurations
│   ├── supabase.ts      # Supabase client
│   ├── prisma.ts        # Prisma client
│   ├── storage.ts       # File storage related
│   └── utils.ts         # Common utilities
└── middleware.ts        # Next.js middleware

📊 Database Models

User

model User {
  id          String      # Unique ID
  email       String      # Email (unique)
  nickname    String      # Nickname
  supabaseId  String      # Supabase user ID
  points      Int         # Points balance
  donations   Donation[]  # Donation history
  createdAt   DateTime    # Created at
  updatedAt   DateTime    # Updated at
}

Donation

model Donation {
  id              String          # Unique ID
  userId          String          # User ID
  imageUrl        String          # Clothing image URL
  brand           String?         # Brand name
  itemType        String          # Item type
  condition       Condition       # Clothing condition (Good/Fair/Poor)
  estimatedPoints Int             # Estimated points
  actualPoints    Int?            # Actual awarded points
  pickupMethod    PickupMethod    # Pickup method (Delivery/Home pickup)
  address         String?         # Pickup address
  status          DonationStatus  # Donation status
  createdAt       DateTime        # Created at
  updatedAt       DateTime        # Updated at
}

🚀 Getting Started

Prerequisites

Node.js 18 or higher
MySQL database
Supabase account (Auth and Storage setup required)
Vercel account (for deployment, free plan available)
AI API service (currently using external service)

Training AI

Download Dataset (refer to dataset description)
Preprocessing
- Please refer to preprocessing.ipynb
Training

yolo detect train data=clothes.yaml model=yolo11n.pt epochs=100 imgsz=640 seed=42 optimizer='Adam' \
     lr0=1e-4 batch=16 cos_lr=True dropout=0.05

Inference

yolo predict model='//PATH/TO/YOUR/TRAINED/MODEL.pt' imgsz=640 conf=0.25

FrontEnd & BackEnd Installation & Setup

Clone the repository

git clone https://github.com/abjin/reward-closet.git
cd reward-closet

Install dependencies

npm install

Environment variables setup Create a .env.local file and add the following:

# Database
DATABASE_URL="mysql://user:password@localhost:3306/reward_closet"

# Supabase
NEXT_PUBLIC_SUPABASE_URL="your-supabase-url"
NEXT_PUBLIC_SUPABASE_ANON_KEY="your-supabase-anon-key"
SUPABASE_SERVICE_ROLE_KEY="your-service-role-key"

# AI API (currently using external service)
# AI_API_URL is hardcoded in the code

Database setup

# Generate Prisma client
npx prisma generate

# Database migration (development)
npx prisma db push

# Or apply migration files (production)
npx prisma migrate deploy

Supabase Storage setup Create the following Storage bucket in your Supabase dashboard:

Bucket name: clothing-images
Public policy: Enabled

Run development server

npm run dev

The application will run at http://localhost:3000.

🌐 Deployment

Vercel Deployment

This project is optimized for deployment on Vercel.

Deploy to Vercel

# Install Vercel CLI
npm i -g vercel

# Deploy project
vercel --prod

Environment variables setup Set the following environment variables in your Vercel dashboard:

DATABASE_URL=your_mysql_database_url
NEXT_PUBLIC_SUPABASE_URL=your_supabase_project_url
NEXT_PUBLIC_SUPABASE_ANON_KEY=your_supabase_anon_key
SUPABASE_SERVICE_ROLE_KEY=your_supabase_service_role_key

Service Architecture

Hosting: Vercel (Next.js optimized)
Authentication: Supabase Auth (social login, email verification)
Storage: Supabase Storage (image file management)
Database: MySQL (Prisma ORM, RDS)
AI Analysis: FastAPI Server on AWS Elastic Beanstalk

📝 Scripts

npm run dev        # Run development server (with Turbopack)
npm run build      # Production build
npm run start      # Start production server
npm run lint       # Run ESLint

🤖 AI Features Details

Dataset

Original Dataset Source: Waste Clothing Recycling Classification and Screening Data (AI-Hub)
Data Preprocessing: Change to YOLO format and integrate data classes (15 classes -> 13 classes) (Details in preprocessing.ipynb)

Training Details

Model: YOLO11n (super light model)
Train/Validation Split: Split train/validation to 8 : 2 (class stratified split)

AI Clothing Analysis API

AI Server Repository: reward-closet-ai-api-server
Framework: FastAPI with PyTorch TorchScript models
Platform: AWS Elastic Beanstalk
API Endpoint: /models/clothes/predict

Supported Categories

Clothing Types (11 categories):

jacket, short pants, tailored pants, jumper, shirts
coat, dress, casual pants, blouse, tshirts, skirt

Defect Detection (5 categories -> 2 categories):

Original: ripped, pollution, tearing, frayed
Ours: tearing, pollution

Condition Classification & Points

Good Condition (500P): No defects detected
Fair Condition (200P): Minor defects (wear, slight damage)
Poor Condition (0P): Major defects (tears, stains, severe damage)

🌟 Main Pages

Home (/) - Service introduction and statistics
AI Prediction (/predict) - Upload clothing photos and AI analysis
Donation Application (/donate) - Apply for donation and pickup information
My Page (/mypage) - Personal donation history and points management
Login/Sign Up (/login, /signup) - User authentication

🎯 How to Use

Sign Up - Simple registration with email and nickname
Upload Photo - Take and upload photos of clothes to donate
AI Analysis - AI analyzes clothing condition and provides estimated points
Apply for Donation - Select pickup method and apply for donation
Earn Points - Receive points after donation completion

🤝 Contributing

Fork the Project
Create your Feature Branch (git checkout -b feature/AmazingFeature)
Commit your Changes (git commit -m 'Add some AmazingFeature')
Push to the Branch (git push origin feature/AmazingFeature)
Open a Pull Request

📄 License

This project is distributed under the MIT License.

📧 Contact

If you have any questions about the project, please create an issue.

Let's create a meaningful donation culture together with Reward Closet! 🌱

Name		Name	Last commit message	Last commit date
Latest commit History 22 Commits
data		data
prisma		prisma
public		public
src		src
.gitignore		.gitignore
README.kor.md		README.kor.md
README.md		README.md
components.json		components.json
diagram.png		diagram.png
eslint.config.mjs		eslint.config.mjs
next.config.ts		next.config.ts
package-lock.json		package-lock.json
package.json		package.json
postcss.config.mjs		postcss.config.mjs
tsconfig.json		tsconfig.json

abjin/reward-closet

Folders and files

Latest commit

History

Repository files navigation