Whereas almost 80% of the world’s knowledge is in video format, enabling search and understanding on video knowledge has traditionally been a difficult, guide and time-consuming endeavor. The crew at Twelve Labs got down to change that by growing multimodal foundational fashions targeted on video — and by constructing a strong platform round these superior fashions to allow enterprises to extract worth from their huge archives of video content material. Throughout that course of, the Twelve Labs crew partnered with MosaicML and Databricks to create their industry-leading fashions and add video intelligence to the Databricks Information Intelligence Platform.
Immediately, we’re excited to announce that Databricks Ventures has invested in Twelve Labs. This funding deepens our longstanding partnership with Twelve Labs and opens the door for added integration factors between Databricks and Twelve Labs to assist unlock the context contained inside video knowledge for a broader set of shoppers.
Why Twelve Labs
Twelve Labs has been on the forefront of fixing probably the most complicated challenges in AI in the present day: video reasoning. Combining deep analysis with real-world buyer suggestions, Twelve Labs has developed a set of proprietary fashions that generate highly effective video embeddings that seize the intricate relationships between visible expressions, physique language, spoken phrases, and total context inside movies. With Twelve Labs, enterprises are actually in a position to effortlessly search and monetize in depth video libraries, extract insights, and repurpose content material at scale.
Twelve Labs’ distinctive research-driven, video-first method has allowed the corporate to land a powerful array of enterprise clients throughout a breadth of video-intensive industries together with sports activities, media and leisure, promoting and extra. Our clients in these industries have more and more been asking Databricks to increase the analytical capabilities of the Databricks Information Intelligence Platform into video knowledge, so discovering methods to work extra carefully with Twelve Labs, the chief in video understanding, was a pure transfer.
What’s Subsequent for Twelve Labs and Databricks
Not too long ago, we introduced an integration between Twelve Labs and Databricks Mosaic AI that reduces improvement time and useful resource wants for superior video purposes. By leveraging multimodal embeddings and environment friendly Vector Search capabilities supplied by Databricks Mosaic AI Vector Search, our integration permits builders to assemble refined video search, suggestion, and evaluation programs.
Sooner or later, we anticipate working with Twelve Labs to deepen our expertise integrations, akin to by enabling entry to Twelve Labs’ video fashions natively throughout the Mosaic AI Mannequin Serving resolution. This could permit our joint clients to way more seamlessly energy superior video understanding use instances — from looking huge quantities of video content material utilizing pure language, to indexing and analyzing video clips and scenes based mostly on textual content or photographs — permitting simpler incorporation of video understanding into current Databricks knowledge pipelines and machine studying workflows. As video content material continues to proliferate throughout industries, from media and leisure to e-commerce and safety, the power to effectively analyze and extract worth from video knowledge will grow to be more and more essential.
We’re thrilled to be working much more carefully with the Twelve Labs crew. Keep tuned for information on future collaboration.
