SIC-Slide-Search

Samsung Innovation Campus Capstone Project

Introduction

This project aims to support students of the Samsung Innovation Campus in easily searching for knowledge within very lengthy slides. This is a project that applies RAG (Retrieval-Augmented Generation) with image data. I use the OpenCLIP model to embed images into vectors and utilize ChromaDB as the vector store.

Result

Result

Result

How to use

You can apply this code for other slides or use cases by following the instruction below

• Store your slides in slides directory
• Install libraries

pip install -r requirements.txt

• Process slides and create vectorstore

python ingest.py

• Run gradio interface

python gradio_app.py

Acknowledge

LangChain Templates: https://templates.langchain.com/