Leading AI and video search company Moments Lab (formerly Newsbridge) has launched an AI research program to advance multimodal and generative AI video understanding with low-energy-consumption models.
France’s leading public graduate school for engineering, Telecom SudParis, and free-to-air broadcaster and streaming platform TF1 Group are the first organizations to join the Moments Lab program, with more research partners to be announced shortly.
The program’s main R&D focus includes long-form video understanding, vision-language models, and information retrieval, with a strong emphasis on building AI indexing models that require minimal energy consumption and learn fair representations of a diverse and multicultural world.
“Content producers and creators already use AI to accelerate workflows and capitalize on their entire media library, largely thanks to video-indexing technologies powered by speech and image processing,” explained Dr. Yannis Tevissen, recently appointed head of science at Moments Lab following the completion of his doctorate in artificial intelligence. “However, the margin for improvement remains huge. Videos are highly complex, with very high dimensional variations. The next frontier for building content from human knowledge likely lies in efficiently understanding what’s inside video files. Doing so at scale with small, energy-efficient models is an even bigger challenge, which Moments Lab is already making headway on.”
Olivier Martinot, director of innovation at Telecom SudParis (IMT Group), said, “Our research partnership with Moments Lab follows the CIFRE thesis led by Dr. Yannis Tevissen and supervised by Telecom SudParis professor Jérôme Boudy. Telecom SudParis is delighted to provide its research skills in the field of content analysis and transform this research into innovation with Moments Lab.”
“What Moments Lab is achieving with its next-gen AI indexing models aligns with TF1’s ambition to evolve our Group and future ways of working with cutting-edge technology,” added Olivier Penin, director of innovation at TF1 Group. “Our goal is to increase productivity and enhance video storytelling through the effective use of AI in media indexing. We know that we will make exciting progress in this area as research partners.”
The research program announcement follows Moments Lab’s multimillion-dollar investment in deep-tech R&D and the successful launch of its game-changing, patented AI indexing technology MXT-1 — which has won several industry awards since its launch in 2023, including the IABM Peter Wayne Golden BaM Award.
The company has launched a new, dedicated research website where it will publish its
findings. More information about Moments Lab’s AI R&D projects can be found at www.research.momentslab.com
Moments Lab will showcase the features of its new MXT-1.5 at the 2024 NAB Show in Las Vegas, April 14-17 in Booth SL2113.
During the 2024 NAB Show, Moments Lab will also be participating in the Programming Everywhere Conference in the Beethoven, Encore Hotel on Sunday, April 14, at 1:45 p.m. PDT. Moments Lab CEO and co-founder Philippe Petitpont will discuss “Building Better Stories Faster with AI Moment-Based Search,” focusing on how multimodal AI indexing technology is transforming production workflows with powerful moments-based search and editorial sequencing.
