Machine vision preserves facial treatment

Despite the fact that data protection concerns have been a factor for many years, it turns out that if you put a useful application in front of the machine vision algorithm – you're entertaining – everyone is happy. For example, a Russian music festival uses a fax sample recognition algorithm for participants with photos of events, while a Singapore company develops a traffic ticket system that develops volunteer face recognition to commiserate fees.

It helps consumers have face detection technology in the palm of their hands. Mobile apps, such as FaceLock, scans the face of a user to unlock apps on their smartphone or tablet. In addition, according to Apple's latest announcement, the new generation of iPhone will use "profound information to enhance face recognition." Users are familiar with face recognition in critical areas such as mobile banking and commerce.

Predicted growth in face recognition and other biometric usage reflects these trends. The Face Detection Market will rise from $ 3.3 billion in 2016 to $ 6.84 billion in 2021. Analysts attribute growth to the growing supervisory market, increasing government deployment and other applications in identity management.

The machine vision industry begins to find the way to face detection growth, whether it's calibrated in a small camera or a mobile app that helps policemen suspect. But the technique must first fight some wrists.

Redact and Serve

Suspect Technologies, a Massachusetts-based company in Cambridge, developed advanced face recognition algorithms, but for two very different purposes in the law enforcement area. One use deals with considerations of private life around body cameras worn by policemen. The most commonly referred to bodyworm video is to improve law enforcement responsibility and transparency. If a request for a Freedom of Information Act is requested to obtain such a video, law enforcement agencies must respond promptly.

But they can not do this without first blurring the identity of victims, minors and innocent performers, which was typically a slow, tedious process that was limited to video professionals. Suspect Technologies & # 39; Automated video redaction (AVR) software available on VIEVU cameras is optimized by BWV in real-world conditions – most of the movement and low illumination. The technology allowing multiple objects to be tracked has a simple interface that allows users to add or adjust new objects. AVR changes the video up to tenfold compared to existing methods.

Contrary to AVR, covering identity, Suspect Technologies launches a mobile phenomena recognition application to identify suspicions. "As it stands, there is no easy way for law enforcement agencies to tell if anyone is a criminal," says Jacob Sniff, CEO and Technical Manager at Suspect Technologies.

Compatible with iPhone and Android devices, the company's cloud-based surveillance software has been tested for 10 million pages. The algorithm uses the right face recognition accuracy, which increases tenfold every four years. "Our goal is to be 100 percent accurate according to 10,000 identities," Sniff says.

Suspect Technologies customizes products to regional law enforcement agencies in medium-sized cities that typically have about 100 earned capability. The company plans to present the software for schools and businesses to participate in applications.

Cameras Recognized

On the hardware side, the Face Detection Application Specification is the choice of a machine vision camera. "Monochrome cameras are more responsive to light sensitivity, so they are ideal for low light indoor and outdoor," says Mike Fussell, marketing manager at the FLIR Systems, Inc. (Wilsonville, Oregon) integrated imaging division. "If someone is heavily backlit or shaded, the latest generation of high performance CMOS sensors really shine in these difficult situations."

FLIR offers customers with better performance in low light conditions for higher levels of sensors with high speeds and global shutter speeds. The full pixel count can be read at once, eliminating the distortion caused by the rolling shutter on less expensive sensors, says Fussell. Roller shutters show distortion caused by the movement of the subject relative to the shutter, but they offer a cheaper alternative to low light conditions.

Most cameras used in Face Detection are in range of 3-5 MP according to Fussell. But in an application such as a passport where all variables are checked, a lower resolution camera is suitable. FLIR also offers stereo visual products calibrated by customers for optical tracking and measuring eye movement relative to the head.

Some companies have led the concept of facial recognition to the next level by walking analysis, studying human motion. "In building automation, where you want to learn the people's habits, you can keep track of how to turn lights on or off, or lifts are waiting for them," says Fussell.

Overcoming obstacles face-to-face

Face-to-face technology faces all the challenges you face before challenging an algorithm to reach your camera or mobile device. According to a study, facial recognition systems are 5 to 10 percent less accurate when trying to identify African Americans as compared to whites. In addition, female subjects are harder to detect than males, and younger subjects are more difficult to identify than adults.

Accordingly, algorithm developers should focus more on the content and quality of training data so that data sets can be shared within demographic data as well. Testing the Face Detection System currently offered by the National Institute of Standards and Technology (NIST) can improve accuracy.

If the algorithm has reached the camera, face recognition depends on the number and quality of the images in the comparative database. And although most fax recognition technologies are automated, most systems require human testing to make the final match. Without vocational training, human surveyors are making a bad decision about the match of half the time.

However, the machine vision industry is not alien to wait for the technology mature. As Face Detection does, camera manufacturers and software vendors are ready to deliver equipment and services for secure and accurate identity verification.

Source by Anish Soneja

Leave a Reply

Your email address will not be published. Required fields are marked *