ClosetAI

🤖 本项目 100% 由 AI 构建，欢迎各位开发者自由使用。如果在使用过程中遇到任何问题，建议直接去问问 AI —— 毕竟它比我更了解这段代码。

An AI-powered wardrobe management iOS app built with SwiftUI. Automatically tag your clothing with AI, create outfit collages, and try on outfits virtually — all on-device with your own API key.

---

📽️ Demo

showcase.mp4

---

✨ Features

衣橱管理（Wardrobe）

• 📸 AI 自动打标 — 拍照 / 相册上传，通义千问 Qwen-VL 自动识别大类、颜色、风格、季节、场合
• 🖼️ 背景消除 + 平铺图 — 自动去背景，生成标准化 flat-lay 展示图
• 🗂️ 分类浏览 — 上装 / 外套 / 连衣裙 / 下装 / 鞋子 / 包包 / 配饰各自独立横栏
• 🔍 全文检索 — 跨颜色、风格、类别、备注的实时搜索
• ✏️ 标签编辑 — Chip 单选大类，多选季节 / 场合，自由编辑小类 / 图案

穿搭创建（Outfit）

• 🎨 AI 穿搭推荐 — 根据场合、季节、数量偏好，AI 从衣橱中智能组搭并生成推荐语
• 🖼️ AI 平铺图生成 — 通义 Wan2.6 生成高质量 flat-lay 穿搭大图
• 👗 虚拟试穿 — 选择模特图 + 穿搭，AI 合成上身效果，支持左右滑动对比
• ✋ 手动组搭 — 手动选取最多 4 件单品，即时组建穿搭方案

数据洞察（Settings）

• 📊 颜色分布 — 可视化衣橱色系占比，区分灰 / 米 / 卡其等细分色名
• 🕐 久未穿着 — 统计 90 天内未出现在穿搭中的单品
• 🔑 API Key 管理 — 本地 Keychain 安全存储，不上传任何服务器

---

🏗️ Architecture

ClosetAI/
├── App/                  # Entry point, ContentView, TabView
├── Models/               # ClothingTags, ClothingItemModel, OutfitModel
├── Persistence/          # CoreData stack, PersistenceController, entity extensions
├── Services/
│   ├── AliyunService.swift      # DashScope API (auto-tag, collage, try-on)
│   └── ImageProcessingService.swift  # Background removal, flat-lay generation
├── ViewModels/
│   ├── WardrobeViewModel.swift  # Wardrobe CRUD, search, statistics
│   └── OutfitViewModel.swift    # Outfit creation, AI calls, save logic
├── Views/
│   ├── Components/       # ViewExtensions (Toast), TagView, AsyncImageView
│   ├── Wardrobe/         # WardrobeView, ClothingDetailView, CameraView
│   ├── Outfit/           # OutfitView, TryOnView, ManualOutfitView
│   └── Settings/         # SettingsView (stats, API key, not-worn items)
├── Utilities/            # AppColors
└── Resources/            # Assets, model_person.png (default try-on model)

Pattern: SwiftUI + MVVM + CoreData Dependency: Kingfisher (remote image loading)

---

📱 Requirements

Item	Requirement
iOS	26.0+ (Xcode 26 beta)
Device	iPhone with camera
API	阿里云 DashScope API Key（免费额度可用）
Tool	XcodeGen (brew install xcodegen)

Note: This project targets iOS 26 and requires Xcode 26 beta. It uses Wan2.6-image and Qwen-VL-Plus models from Alibaba Cloud DashScope.

---

🚀 Getting Started

1. Clone & Generate Project

git clone https://github.com/<your-username>/ClosetAI.git
cd ClosetAI

# Install XcodeGen if needed
brew install xcodegen

# Generate the Xcode project
xcodegen generate

2. Configure Signing

Edit project.yml and fill in your Apple Developer Team ID:

settings:
  base:
    PRODUCT_BUNDLE_IDENTIFIER: com.yourname.closetai  # Change as needed
    DEVELOPMENT_TEAM: "XXXXXXXXXX"                    # Your Apple Team ID

Then regenerate:

xcodegen generate

3. Open in Xcode

open ClosetAI.xcodeproj

Select your device and hit Run (⌘R).

4. Add Your DashScope API Key

On first launch, go to 设置 (Settings) tab → enter your DashScope API Key. The key is stored securely in iOS Keychain and never leaves your device.

Get a free API key at: https://dashscope.aliyun.com/

---

🤖 AI Models Used

Function	Model	API
服装自动打标	qwen-vl-plus	DashScope OpenAI-compatible endpoint
AI 穿搭平铺图	wan2.6-image	DashScope multimodal generation
虚拟试穿	wan2.6-image	DashScope multimodal generation

All AI calls are made directly from the app to DashScope using your personal API key. No intermediate server involved.

---

🔒 Privacy & Security

• No data collected — all clothing photos and outfit data are stored locally in CoreData / app Documents directory
• API key in Keychain — stored using iOS Security framework, never hardcoded or logged
• No analytics or tracking of any kind
• Photos are sent to DashScope API only when you trigger AI features (tagging / outfit generation / try-on)

---

🗺️ Roadmap

• 穿着日历 — visual calendar of past outfits
• iCloud / OSS 云同步 — cross-device wardrobe sync
• 季节智能推荐 — weather-aware outfit suggestions
• Widget — daily outfit reminder on home screen
• CoreData → CloudKit migration

---

🤝 Contributing

Pull requests are welcome! For major changes, please open an issue first to discuss what you'd like to change.

Setup for contributors:

1. Follow the "Getting Started" steps above
2. Use swiftc -parse <file.swift> for quick syntax checks without a full build
3. The project uses XcodeGen — edit project.yml for project structure changes, then run xcodegen generate

---

📄 License

MIT

---

📚 Technical Documentation

See TECH_REPORT_v2.0.md for detailed architecture notes, CoreData schema, and build instructions.