Samsung backs a video AI startup that can analyze thousands of hours of footage

Many AI tools can focus on a video today and shortens what is going on but when you ask models about multiple videos and multiple videos extensively multiple videos and footage, the topic becomes a bit complicated.

This is a major limitations for security agencies that want to use AI through thousands of hours of footage from different cameras, as well as marketing companies who want to study various videos and shoot products.

Memory.E. It wants to deal with the AI platform that can process the video for up to 10 million hours. For companies to analyze lots of videos, startup wants to provide a relevant layer, complete with searching index, tagging, categories and sums.

Its co-founder, Dr. Shawn Shane, was a research scientist in the Materi Reality Labs while doing his PhD, and his opponent, Anmin (Ben) Jho Meta, worked as a machine learning engineer.

“All the top AI companies, such as Google, OpenAI and Meta, focus on producing models from end to end

“But when people use visual memory, we flow into a large context of the data. We were inspired by it and wanted to create a solution to understand the video for many more hours,” he said.

To that goal, the company has now collected millions of dollars in a seed funding round, led by Susa Venture, and with the participation of Samsung Next, Fusion Fund, Crane Ventures, Cedcamps and Creator Ventner. Shane said the company had initially aimed at raising $ 1 million, but the investors’ interest ended with the overseas round.

“Shane is a very technical founder, and he is interested in understanding video understanding and intelligence boundaries,” said Misha Gordon-R, partner of Susa Venture. “Memories.This can unlock first -party visual intelligence data with his solution.

Samsung had some different thesis next: Samsung’s investment forces are useful for customers to see the memory of the IAI IA solution.

“We’ve liked a thing about memories i

Memories.AI says that it uses its own technology stacks and models to perform analyzes. First, it removes the sound from the videos and only passes the output with an abbreviation layer to store important data. Then there is an index level, which makes video data searchable (using natural language questions) with categories and tags. There is also a sum level that shortens data from the index, helping to create reports.

Currently, startup provides two types of companies: marketing and protection. Marketing companies can use startup tools to find trends related to their brands on social media and to identify what kind of videos they want to create. Memories.The marketers provide equipment to make those videos.

People are also working with security agencies with security agencies to determine the potential dangerous action by logic through patterns.

Currently, companies working with the monuments will need to upload their video library platform to analyze IA clips. However, Shane said that in the future, its clients will be able to easily create a share drive and sync more easily. The plan is to enable customers to ask this national question: “Tell me all about the interviews I took last week.”

Shane imagines an AI assistant that can achieve the context through their photos in the context of a user’s life or when they activate smart glasses. He sees Humanoid robots to play a role in doing complex tasks or helping self-driving cars to help in remembrance of different routes.

The company currently appoints 15 people and it is planned to use the fund to increase its team and improve its search.

Memories.This is rising against these similar startups, as Mem0 And ReadWhich is working to provide a memory level for AI models, though they provide limited video support at the moment. It also has to fight with companies like Toelvelab and Google, which is working to help AI models understand videos.

Shane, of course, thinks that the solution to his company is even more horizontal, which allows it to work with various video models.