Samsung supports the start of AI video starting to analyze thousands of hours from the shots

Many artificial intelligence tools can look at a video today and summarize what is going on, but things become somewhat difficult when they ask questions on models about many videos and shots that extend for long hours.

This is the great restriction of security companies that want to use artificial intelligence for cleaning during thousands of hours of shots from different cameras, as well as marketing companies that want to study various video campaigns and shoot products.

Memories He wants to address this problem with the artificial intelligence platform that can process up to 10 million hours of video. For companies that have a lot of video to analyze, you want to start off a contextual layer, with a searler, brand, sectors and assembly indexing.

The co -founder of Dr. Seen Shin was a research scientist in the laboratories of Meta while he was getting a doctorate, and his counterpart INGen (Ben) Zhu worked in Meta as an engineering engineer.

“All the first AI companies, such as Google, Openai and Meta, focus on producing comprehensive models. These capabilities are good, but these models often have restrictions on understanding the video context until an hour or two.”

“But when humans use visual memory, we get to know a great context of the data. We are inspired and we wanted to build a solution to understand the video within a better long hours,” he said.

To achieve this goal, the company now raised $ 8 million in the seed financing round led by Susa Ventures, and the participation of Samsung Next, Fusion Fund, Crane Ventures, Seedcamp and Creator Ventures. Shin said that the company initially aims to raise 4 million dollars, but it ended with an excessive production tour due to the benefit of the investor.

“Shin is a very technical founder, and he is obsessed with pushing the limits of understanding video and intelligence,” said Misha Gordon Row, a partner in Sosa Ventures. “Memories. A lot of visual intelligence data can be locked at the first end,” he added.

Samsung had a slightly different thesis: the investment arm in Samsung sees memories. AII solution is useful for consumers.

“One of the things we liked in memories. AII is that you can do a lot of computing on devices. This means that you do not necessarily need to store video data in the cloud. This can open better security applications for people who are afraid to put security cameras in their homes due to privacy concerns.”

Memories.ai says it uses its technology staple and its models to perform analyzes. First, it removes noise from videos and passes the output through a pressure layer to store important data only. Then there is an indexing layer, which makes video data (using natural language queries) with fragmentation and marks. There is also a collection layer that summarizes data from the index, which helps to create reports.

Currently, it meets the startup to two types of companies: marketing and security. Marketing companies can use startup tools to search for trends related to their commercial brands on social media, and determine the type of video they want to make. Memories.ai also provides tools for marketing to create these videos.

The company is also working with security companies to help them analyze safety footage to determine the procedures that can be dangerous by people in videos by thinking through patterns.

Currently, companies that work with memories.ai need to download their video library to the articles of association to make clips analysis. But Shin said that in the future, his client will be able to create a shared engine and synchronize more easily. The plan is to enable customers to ask questions such as: “Tell me everything about the people I met last week.”

Amnesty International Assistant Chene can gain a context of the user’s life through his photos or when they activate smart glasses. He also believes that technology plays a role in training human robots to do complex tasks or help self -driving cars in remembering different ways.

The company is currently employing 15 people, and plans to use the fund to increase its team and improve its research.

Memories.ai escalating against similar startups, such as Mem0 and ReadThat provides a memory layer for artificial intelligence models, although it provides limited video support at the present time. It also has to face companies such as Twelvelabs and Google, which help artificial intelligence models to understand videos.

However, Shin feels that his company solution is more horizontal, which will allow it to work with various video models as well.