AI has broadened the paths beyond facial & voice recognition, by embedding the analytics in biometrics in law enforcement to pace the surveillance process.
Artificial intelligence is alleviating the security & law enforcement of Biometrics by providing digital monitoring in the surveillance by the video analytics, LiDAR technology & Mining the data from the prisons.
According to the Wikipedia,
Biometrics are measurements & calculations of the body linked with the human characteristics. Authentication in Biometrics uses computer science in the form of identification & access control. It also uses the individual identification in groups under the surveillance.
The biometric is originated back in 1981 to identify the fingerprints of criminals which is analyzed & stored. Now it is used in various other fields such as Financial Services, National and political activities, Advertising & psychological researchers, & Behaviometrics.
Biometric was initially used in the access control or validating transactions, but now it has broadened the paths of authentication using iris scanning, fingerprint scanning, facial & voice recognition. For authentication, they use biological characteristics like chemistry, human physiology, or behavior.
The Use Cases of Biometric Authentication:
- criminal identification
- enemy identification
- passenger identification
- voter identification
- patient identification
- employee identification
- customer identification
How AI is working in Biometrics:
There are two identification methods the AI is using in the Biometric: Physical Biometrics & Behavioral Biometrics.
It uses human characteristics such as a person's face, DNA, iris, & fingerprints and transforms this information into an understandable AI code language. It is further divided into recognition of Ear, Eye vein, Facial, Fingerprint, DNA matching & Footprint, Keystroke (rhythm with which one types at a keyboard), and foot.
It operates in the same way as the physical but uses human behavior such as a person's typing rhythm, the way a person interacts with the devices, voice, to analyze the characteristics. It is the scientific study of how the human body function. It is divided into three categories: Signature, rhythm & voice.
Advancements in Biometrics:
1.) SVI and IDEMIA NSS LiDAR Biometric:
SVI(StereoVision Imaging, Inc.) is working on the 2D/3D/4D LiDAR-based facial/object recognition/remote sensing technology. It has tied up with the IDEMIA National Security Solutions(NSS) to deploy the LiDAR biometric recognition to law enforcement.
LiDAR is the remote sensing technology used for measuring the distance of the object on the earth's surface. There are three primary components in the LiDAR is the GPS receiver, laser & scanner. The principle is simple, throw a laser light at an object on the earth’s surface and calculate the time it takes to return to the LiDAR source. SVI laser technology can seize the velocity or motion information in real-time, subject's movements, lighting conditions, it can detect the heart and breath rate, eye movement & speech.
IDEMIA is helping French law enforcement to fight against crime & illegal immigration:
French data decided to innovate the passenger data management to detect the threats & reinforcing homeland security by deploying IDEMIA's end-to-end API solutions. As air travel continues to surge, it has become crucial to strengthen security. To help the French government, IDEMIA offered a sophisticated system to design an automated risk assessment to detect the person of interest and identify the suspicious patterns.
2.) Mining the Data from Prison Phone calls:
Prison phone calls have employed speech recognition, machine learning software & semantic analysis to build the databases of particular texts and comprehension. US prisons and jails have collaborated with the LEO AI-driven company, which mines the inmate recorded calls in the database and monitors their inmate population.
LEO technology drives its own investigators to direct the investigators to seed the database with their data, which is prison slang or a particular phrase. And if the system picks up some suspicious language or any phrase, they respond to the law enforcement. The same process goes in monitoring the recorded conversations.
Inside the jail, criminal conduct is conveyed in the codes and where words can have a different meaning also. For instance, "marijuana" is sometimes called the 'spice' and 'tube or 'stick' is also called the 'mower', all slangs will be fed in the database to identify any suspicious activities.
3.) Identifying the suspects in public spaces with IRIS-Unleash AI Video Analytics:
Law enforcement is stretching their searching levels to find the criminals and secure the public spaces from suspicious activities by introducing video analytics through the IRIS-UNLEASH platform, The ArcGIS IRIS is cloud-based GIS mapping software that connects people, locations, and data using interactive maps and Unleash is a live media platform which captures through the aerial camera drones and sensors and other robotic cameras to the cloud for AI-powered video & image processing
Unleash integrates the output of the CCTV camera networks, existing IoT devices, and other data sources to capture the dynamic data extracted from the video feeds from every corner of the cities to identify, inform, predict and alert. From law enforcement to governments, city planners to public works, all are helped immensely by this technology.
Facial and license plate recognition can be used to detect people and particular vehicles in real-time through video surveillance like CCTV cameras, traffic cameras or online video feeds. For instance, you can detect the suspect or unauthorized person.
The basic tasks of the video analytics:
- Image classification: the image can be a car or person
- Localization: Locate an image in the video
- Object detection: Identify the object
- Object tracking: track an object in the video
To mitigate the Covid-19, the technology is advancing the features in the facial recognition:
Facial recognition identifies and analyzes the person and verifies its identity by comparing the digital image or video frame to the face stored in the database.
What if the face is covered?
As we are in the mid of the COVID-19, the tech has to advance the facial recognition because half of the faces are covered with face masks. But the technology is mitigating this problem too, the traditional facial recognition logic revolves around identifying the several key points of the face, but when the bottom part is hidden, many of the points get missed out. To enable successful recognition, the eyes and nose are important.
On May 19, analysts from the University of Bradford has achieved a 90% identification success rate with the bottom face only. The SenseTime of China pioneered in the real world by identifying the 240 reference points on the individual's face. Apple is thinking out of the box to improve the face feature. In July it got the patent as the "difficult biometric authentication cases". Because the blood vessels underneath the skin of the face is unique for the individual, this type of imaging leaves no room for error and has taken machine learning to the next level of advancements.