Upload photos or take them instantly to find similar clothes | Try on how clothes look on you before buying them |
Describe what you're looking for and we'll find it | Intelligent detection of garments in your photos |
graph TD
A[User]-->|Uploads photo/audio| B[Web]
B -->|Image| C[Cloth Segmentation Mini]
B -->|Audio| D[Whisper-Query-Parser]
C -->|Segmentation| E[Inditex Visual Search API]
D -->|Structured query| F[Inditex Product Search API]
E -->|Similar garments| B
F -->|Search results| B
B -->|Image + Garment| G[Virtual Try-On API]
G -->|Generated image| B
B -->|Shows results| A
Clothing segmentation module that uses a U2NET model to identify and classify garments into:
- 🔴 Upper clothing (class 1)
- 🟢 Lower clothing (class 2)
- 🔵 Full body clothing (class 3)
Converts voice commands into structured queries:
- 🎙️ Audio to text transcription (OpenAI Whisper)
- 🧠 Generation of structured queries (Google Gemini)
- 🖼️ Contextualization with images
Allows virtual garment try-on:
- 📱 Simple Streamlit interface
- 🎯 Realistic visualization
- ⚡ Fast processing with Fashn.ai API
# Install FFmpeg for audio processing
# Ubuntu/Debian
sudo apt-get install ffmpeg
# macOS
brew install ffmpeg
# Windows
# Download from https://ffmpeg.org/download.html
- Clone repositorie
git clone https://github.com/tu-usuario/fashionfinder.git
cd fashionfinder
- Install dependencies
pip install -r requirements.txt
- Config env
cp .env.example .env
# Edita .env con tus claves API:
# - GOOGLE_API_KEY para Gemini
# - FASHN_API_KEY para el try-on virtual
- Start services
-
Follow the guides in the respective READMEs
TODO: COMO CORRER WEB
- Upload a photo or take one with the camera
- The image will be automatically segmented
- Click on any detected garment to see similar options
- Click on the microphone button
- Describe the garment you're looking for (e.g., "I want a blue short-sleeve t-shirt")
- The system will translate your request and show relevant results
- Select an image of yourself and a garment
- Click on "Try-On"
- Visualize how the garment would look on you without needing to try it on physically!
This proyect is licensed under the MIT License - see the LICENSE file for more details.
Try our special feature, take photos or upload your own photos to the app and watch the magic happen!