Build a SQLite vector database for image search using embedding representations and cosine similarity. Learn to create a multi-modal image search system with the latest text embeddings. Explore techniques for parallel processing of GPT Vision descriptions for 500 images, generating embeddings efficiently, and implementing both terminal and web-based image search interfaces. Gain hands-on experience with SQLite, OpenAI API, GPT-4 Vision, and vector operations in Python to create a powerful image retrieval system.
Building a Local SQLite Vector Database for Multi-Modal Image Search
Overview
Syllabus
DEMO
database
SQLite db
Getting descriptions with GPT Vision for 500 images in parallel
echohive patreon
Getting descriptions with GPT Vision continued
Getting embeddings in parallel
Terminal image search
Image search with webpage
Taught by
echohive
