Intel developed new video search technology

Intel is researching a new video search technology that it hopes will turn it into a future multimedia platform.

This technology is being developed by Intel labs in the US and China. It cuts videos into serial frames and then uses images and facial recognition technology to recognize faces, objects, speech, location and movement. The frames will then be joined together to make the video searchable.

For example, users will search for videos about football as soon as their favorite players score. The technology will also identify and classify players' faces and objects such as balls and frames using algorithms and statistical processing technology developed by Intel.

When the user issues a goal review query, the technology will search for frames containing related objects and return the video to the user.

Users can also view only a certain clip without having to watch the entire video. Technology identification can categorize images according to people and objects, helping users not to type keywords about classification tags (tags).

However, according to Lin Chao, an Intel researcher, new technology also has challenges to overcome. Video processing can make searching take hours if the PC's current processing capability is not increased. Chao is also unable to predict when new technology will come to users.

(Photo: vnmedia.vn)

Video search technology is part of Intel's "virtual computing" strategy to connect multiple chip cores, software development platforms and the ability to process graphics together to create better communication between people and PC. Intel wants to use virtual computing platform to bring communication to the PC in 3D-like environments or capable of continuous video analysis.

Intel is also working on a platform called Larrabee, which connects multiple processor cores, scripts and graphics processing capabilities to provide the ability to handle up to trillions of calculations per second. Larrabee is expected to be announced in 2010.

Intel researchers are also promoting a project capable of tracking human behavior for nannies. The company has launched a pilot project in Seattle, where 20 families have installed screens to monitor human behavior.

RFID tags are attached to objects such as toothbrushes, combs, vials .; and when these objects are moved, the tag tells the electronic monitor screen, and other family members can know who moved the object.

Currently behavioral monitoring technology of this type has not been implemented because their accuracy is only 70-90%. According to Matthai Philipose, an Intel researcher, this level must be 95 - 98% to meet the requirement. According to him, there is currently no research on this technology, and that their commercialization may not become a reality in the future.