Build AI Apps with Ease with the Milvus Notebook Gallery & Advanced Video Search with Twelve Labs and Milvus!

In this issue:

• New! Milvus Notebook Gallery
• Community Spotlight?
• Milvus X Twelve Labs?
• Upcoming Events?

??Milvus Notebook Gallery?

Build AI Apps with Ease - get started with our interactive notebooks that guide you effortlessly through building cutting-edge GenAI applications.

Find them easily through our NEW Notebook Gallery with tutorials on hybrid search, RAG with Milvus and LlamaIndex, multimodal RAG, and more!??

Get Started: Build AI Apps with Milvus: Tutorials & Notebooks ?

?? Advanced Video Search: Milvus X Twelve Labs

?? Exciting News in AI Video Understanding: Twelve Labs & Milvus Join Forces!

Twelve Labs and Milvus have partnered to support video search applications. This collaboration combines Twelve Labs' multimodal video embeddings with Milvus' efficient vector database, powering new use cases for developers and businesses.

?? Why This Matters:

1. Advanced Semantic Retrieval: Twelve Labs' Embed API + Milvus' vector database = developers can create powerful semantic search engines for video content.
2. Scalability: Milvus offers a scalable solution for managing and querying vast amounts of vector embeddings, perfect for large video libraries.
3. Multimodal Understanding: Twelve Labs' video foundation models provide comprehensive embeddings that capture visual, audio, and textual information from videos.
4. Customization: The integration allows for clip-level or video-level embedding generation, offering flexibility for various use cases.

?? Who Can Benefit:

• Content Platforms: Enhance video discovery and recommendations.
• E-Learning Providers: Improve searchability of educational video content.
• Media Companies: Streamline content management and retrieval.
• Researchers: Facilitate advanced video analysis and retrieval.

??? Getting Started: Interested in using this powerful integration? Check out the resources:

• Twelve Labs API Documentation
• Milvus Documentation
• Sample implementation code for semantic retrieval

This integration marks a significant step forward in making video content as searchable and analyzable as text. Whether you're building a search engine, a recommendation system, or diving into content-based video retrieval, the Twelve Labs and Milvus integration provides the tools you need to bring your ideas to life. ?

Read more: Twelve Labs and Milvus for Semantic Retrieval

Talk by James Le

?? Community Spotlight: Reverse Image Search

Abdullah Abid Ansari created a cool AI reverse image search tool for Wallpaper Engine, a popular Steam app for animated wallpapers. This tool can search through 2.4 million images in just 30-50ms, showcasing the power of modern AI and database technologies.

?Key highlights from the project:?

Problem Solved: Addressed a community need for a reverse search function in Wallpaper Engine's extensive content library.

Data Collection: Extracted metadata for 2.4 million listings from Steam Workshop using Steamworks Web API.

AI Integration: Utilized Facebook's DINOv2-Base ViT model to generate 768-dimensional vector embeddings for each image.

Processing Power: Leveraged dual T4 GPUs on Kaggle for 3 days of continuous computation, processing 982GB of downloads.

Database Evolution: Initially used MongoDB Atlas but faced challenges with storage efficiency and search speed.

Milvus Implementation: Transitioned to Milvus, a high-performance vector database, using Microsoft's DiskANN vector index for sub-100ms search times.

Tech Stack: Employed FastAPI for the backend, NGINX, Docker for database management, and React for the frontend.

LinkedIn post

Try it out: Wallpaper Engine Reverse Image Search ?

-

Find a problem. Solve it with a vector database. Tag Milvus to be featured. ??

??? Upcoming Events

October 4: Grace Hopper Conference - Open Source Day (virtual)?

Stefan Webb and the team will be presenting Milvus at Open Source Day this Friday! Join us for the morning or afternoon session to learn what a vector database is and what cool stuff you can build with them ??

Register: Grace Hopper Celebration ?

October 7: Devoxx Genesis in Belgium (in-person)?

Find Stephen Batifol on Monday at Devoxx Genesis in Belgium speaking about? how to build an advanced Agentic Retrieval-Augmented Generation (RAG) system using entirely open-source technologies.?

The talk will include a live demo, showcasing the end-to-end workflow and performance of the Agentic RAG system.

Register: Devoxx Belgium ?

Stephen’s talk: Build an Agentic RAG system using Langchain, Ollama and Milvus ?

October 8: SF #TechWeek Data & AI Edition (in-person)?

Zilliz is excited to collaborate with BentoML , Timeplus , Datastrato and Tecton on a #TechWeek event on October 8 at SAP in San Francisco! There will be food, networking, and raffle prizes for all! Join us for exciting talks:

?? Multimodal Search with Open-Source Tools, Zilliz

?? A Guide to Compound AI Systems, BentoML

?? One single binary to tackle streaming and historical analytics, Timeplus

Save Your Spot ???

Oct 10: Multimodal Pipelines for AI Apps: Journey To Day 2 (virtual)

Discover how to navigate the journey from GenAI prototypes to production-ready AI applications with Sam Lachterman . Learn about crucial operational concerns like scalability, governance, and automation in AI pipelines. Explore Datavalo, powered by Apache NiFi, and gain insights on implementing production-grade multimodal pipelines for AI, essential for data engineers in the evolving landscape of enterprise AI integration.

Save Your Spot

-

Come back next week for more! ??