FashionAI Documentation

FashionAI is an AI-powered clothing detection system that uses YOLOv8 deep learning to detect and classify clothing items in images. It can identify 8 different types of clothing and classify their colors into 8 color categories.

The system is built with a modern tech stack including PyTorch for deep learning, FastAPI for the backend API, and a responsive web interface for easy interaction.

1. Install Dependencies

Navigate to the backend directory and install required packages:

cd backend
pip install -r requirements.txt

2. Add Models

Place the trained models in the models/ folder:

• cloth_classifier.pt - Clothing detection model
• color_classifier.pt - Color classification model

3. Run Backend

cd backend
uvicorn app.main:app --reload --port 8000

4. Open Frontend

cd frontend
python -m http.server 5500

Then open http://127.0.0.1:5500 in your browser.

Image Upload

1. Go to the App page
2. Click on the upload area or drag and drop an image
3. Adjust the confidence threshold if needed (default: 25%)
4. Click "Detect Clothing" button
5. View results with bounding boxes and classifications

Live Camera

1. Go to the App page
2. Click the "Live Camera" tab
3. Allow camera access when prompted
4. Choose between Capture Mode (take photo) or Live Detection (real-time)
5. View detection results in real-time

The model can detect and classify the following 8 clothing types:

Each detected clothing item is also classified by its dominant color:

The backend exposes a REST API for clothing detection:

Example Response

{
  "success": true,
  "detections": [
    {
      "className": "Tshirt",
      "confidence": 0.95,
      "bbox": [100, 100, 200, 200],
      "color": "blue",
      "colorConfidence": 0.87,
      "colorHex": "#3B82F6",
      "label": "Blue Tshirt"
    }
  ]
}

HuggingFace Spaces

Deploy to HuggingFace Spaces using Gradio:

1. Create a new Space with Gradio SDK
2. Upload files from deployment/huggingface/
3. Upload both model files (.pt)
4. Your app will be live automatically!

Docker

Build and run with Docker:

docker build -t fashionai .
docker run -p 8000:8000 fashionai