This project is a Python-based application that uses LangChain, LangChain-Ollama, and the Serper API to analyze an image of a clothing item, describe it, generate a search query, and recommend online stores where similar items can be purchased. The application leverages local vision and text language models (via Ollama) and web search capabilities to provide a seamless clothing search experience.
- Image Analysis: Uses a vision model (
llava:13b) to describe clothing items in an uploaded image. - Search Query Generation: Generates concise search queries (up to three words) based on the image description using a text model (
qwen2.5). - Web Search Integration: Fetches relevant search results from the web using the Serper API.
- Store Recommendations: Recommends up to five online stores with links to purchase similar clothing items.
- Workflow Orchestration: Utilizes LangGraph to manage the multi-step process of image processing, query generation, searching, and recommending.
Before running the application, ensure you have the following:
- Python 3.8+ installed.
- Ollama installed and running locally with the following models:
qwen2.5:latest(text model)llava:13b(vision model)
- A Serper API key for web search functionality.
- A GPU is recommended for faster processing with Ollama models.
- An image file (e.g.,
red_skirt2.jpg) in the project directory for testing.
- Clone the Repository:
git clone https://github.com/armanjscript/Clothing-Search-Application.git cd Clothing-Search-Application