Electronic Components News & Application
Infineon / Mitsubishi / Fuji / Semikron / IXYS / Eupec

How do sensors play into AI’s facial recognition capabilities? Part 1

Posted Date:2023-09-06 18:14:12

Among the growing number of applications for artificial intelligence (AI) is advanced imaging capabilities for reliable facial recognition. Reliability is required for effective analysis by law enforcement agencies, access to commercial or industrial complexes, travel, or even a homeowner. While AI enables complex analysis and performs it in a remarkably short timeframe, effective analysis requires the best capability from the first block in the system, the sensor – the digital camera.

While they had video evidence, movies and television shows have often complicated the apprehension of a suspect based on the resolution (number of pixels) of the camera being too low – usually based on the owner choosing the cheapest system available. Today’s high-end cameras are vastly improved and even the cheaper versions have much improved capabilities over the past two decades.

The best possible camera resolution available today for high-definition security cameras is 8K. However, cameras with this resolution setting require camera servers to handle the amount of data. In contrast, 4K cameras are more practical for many applications when performance and camera resolution are the focus.

Unlike a 4 mega-pixel (MP) camera that has 2560 x 1440 or 3,686,400 total pixels, a 4K rating provides 3840 x 2160 or 8,294,400 total pixels, over twice as many pixels. For further comparison, the 4K’s pixels are a little more than 4x the size of a commonly rated 1080P (1920 x 1080 or 2,073,600 total pixels) display.

With facial recognition, each observed face is compared against a database creating a much more advanced system. This creates two distinct classifications for home and other security cameras: cloud based – those that send facial images to the manufacturers’ servers to be stored and identified in the cloud and premise based – those that run the facial recognition algorithms on the camera itself and do not send the images to the manufacturer.

In all security systems, one of the key benefits of facial recognition is minimizing the number of false alerts.In addition, by providing useful information and statistics about specific people, facial recognition cameras cut down the time the system and analysts spend tracking activity.

In the home (a house or apartment), facial recognition is increasingly used in home security cameras and video doorbells. According to Future Marketing Insights, one of the reasons that sales of smart home security cameras will grow at a rate of 20.02% from 2022 to 2032 is more secure deliveries. The report states, “With the integration of artificial intelligence in smart cameras, owners get immediate notification of deliveries.”

But the use of interior cameras will also grow based on monitoring and even communicating with children and pets through smart cameras when the homeowner is not at home.

In addition to home security systems, advanced cameras with AI are being designed to monitor activity inside schools, offices, restaurants, bars, and malls to address the increasing threats that bad actors pose to the intended users. These are key areas where avoiding false flags are essential, so an immediate response can occur when a true red flag is raised.

Part 2 of this blog will address addition applications and use cases including the Transportation Security Agency’s (TSA’s) use of facial recognition technology in airports across the United States.

CCTV Resolution Guide – A1 Security Cameras
Facial Recognition Come to Your Neighborhood Through Security Cameras and Video Doorbells – Consumer Reports

Source from: https://www.shunlongwei.com/how-do-sensors-play-into-ais-facial-recognition-capabilities-part-1/

category: Components

Pre:How do sensors play into AI’s facial recognition capabilities? Part 2

Next: If you are working with sensors here are some tools to consider: Part 1